markdown
stringlengths 0
1.02M
| code
stringlengths 0
832k
| output
stringlengths 0
1.02M
| license
stringlengths 3
36
| path
stringlengths 6
265
| repo_name
stringlengths 6
127
|
|---|---|---|---|---|---|
4.3 Label Distribution
|
print(train_df["label"].value_counts())
fig = plt.figure(figsize=(10, 6))
label_stats_plot = train_df["label"].value_counts().plot.bar()
plt.tight_layout(pad=1)
plt.savefig("img/label_stats_plot.png", dpi=100)
|
half-true 2114
false 1995
mostly-true 1962
true 1676
barely-true 1654
pants-fire 839
Name: label, dtype: int64
|
MIT
|
notebooks/eda-notebook.ipynb
|
archity/fake-news
|
4.4 Speaker Distribution
|
print(train_df.speaker.value_counts())
fig = plt.figure(figsize=(10, 6))
speaker_stats_plot = train_df["speaker"].value_counts()[:10].plot.bar()
plt.tight_layout(pad=1)
plt.title("Speakers")
plt.savefig("img/speaker_stats_plot.png", dpi=100)
print(train_df.speaker_title.value_counts())
fig = plt.figure(figsize=(10, 6))
speaker_title_stats_plot = train_df["speaker_title"].value_counts()[:10].plot.bar()
plt.tight_layout(pad=1)
plt.title("Speaker Title")
plt.savefig("img/speaker_title_stats_plot.png", dpi=100)
|
President 492
U.S. Senator 479
Governor 391
President-Elect 273
U.S. senator 263
...
Pundit and communications consultant 1
Harrisonburg city councilman 1
Theme park company 1
Executive director, NARAL Pro-Choice Virginia 1
President, The Whitman Strategy Group 1
Name: speaker_title, Length: 1184, dtype: int64
|
MIT
|
notebooks/eda-notebook.ipynb
|
archity/fake-news
|
4.5 Democrats vs Republicans* Let's see how the 2 main parties compete with each other in terms oftruthfulness in the labels
|
fig = plt.figure(figsize=(8,4))
plt.suptitle("Party-wise Label")
ax1 = fig.add_subplot(121)
party_wise = train_df[train_df["party_affiliation"]=="democrat"]["label"].value_counts().to_frame()
ax1.pie(party_wise["label"], labels=party_wise.index, autopct='%1.1f%%',
startangle=90)
ax1.set_title("Democrat")
plt.suptitle("Party-wise Label")
ax2 = fig.add_subplot(122)
party_wise = train_df[train_df["party_affiliation"]=="republican"]["label"].value_counts().to_frame()
ax2.pie(party_wise["label"], labels=party_wise.index, autopct='%1.1f%%',
startangle=90)
ax2.set_title("Republican")
plt.tight_layout()
plt.savefig("img/dems_gop_label_plot.png", dpi=200)
|
_____no_output_____
|
MIT
|
notebooks/eda-notebook.ipynb
|
archity/fake-news
|
* We can combine some labels to get a more simplified plot
|
def get_binary_label(label):
if label in ["pants-fire", "barely-true", "false"]:
return False
elif label in ["true", "half-true", "mostly-true"]:
return True
train_df["binary_label"] = train_df.label.apply(get_binary_label)
fig = plt.figure(figsize=(8,4))
plt.suptitle("Party-wise Label")
ax1 = fig.add_subplot(121)
party_wise = train_df[train_df["party_affiliation"]=="democrat"]["binary_label"].value_counts().to_frame()
ax1.pie(party_wise["binary_label"], labels=party_wise.index, autopct='%1.1f%%',
startangle=90)
ax1.set_title("Democrat")
plt.suptitle("Party-wise Label")
ax2 = fig.add_subplot(122)
party_wise = train_df[train_df["party_affiliation"]=="republican"]["binary_label"].value_counts().to_frame()
ax2.pie(party_wise["binary_label"], labels=party_wise.index, autopct='%1.1f%%',
startangle=90)
ax2.set_title("Republican")
plt.tight_layout()
plt.savefig("img/dems_gop_binary_label_plot.png", dpi=200)
|
_____no_output_____
|
MIT
|
notebooks/eda-notebook.ipynb
|
archity/fake-news
|
5. Sentiment Analysis
|
from textblob import TextBlob
pol = lambda x: TextBlob(x).sentiment.polarity
sub = lambda x: TextBlob(x).sentiment.subjectivity
train_df['polarity_true'] = train_df[train_df["binary_label"]==True]['statement'].apply(pol)
train_df['subjectivity_true'] = train_df[train_df["binary_label"]==True]['statement'].apply(sub)
plt.rcParams['figure.figsize'] = [10, 8]
x = train_df["polarity_true"]
y = train_df["subjectivity_true"]
plt.scatter(x, y, color='blue')
plt.title('Sentiment Analysis', fontsize=20)
plt.xlabel('<-- Negative ---------------- Positive -->', fontsize=10)
plt.ylabel('<-- Facts ---------------- Opinions -->', fontsize=10)
plt.savefig("img/sa_true.png", format="png", dpi=200)
plt.show()
train_df['polarity_false'] = train_df[train_df["binary_label"]==False]['statement'].apply(pol)
train_df['subjectivity_false'] = train_df[train_df["binary_label"]==False]['statement'].apply(sub)
plt.rcParams['figure.figsize'] = [10, 8]
x = train_df["polarity_false"]
y = train_df["subjectivity_false"]
plt.scatter(x, y, color='blue')
plt.title('Sentiment Analysis', fontsize=20)
plt.xlabel('<-- Negative ---------------- Positive -->', fontsize=10)
plt.ylabel('<-- Facts ---------------- Opinions -->', fontsize=10)
plt.savefig("img/sa_false.png", format="png", dpi=200)
plt.show()
|
_____no_output_____
|
MIT
|
notebooks/eda-notebook.ipynb
|
archity/fake-news
|
Collaborative filtering> Tools to quickly get the data and train models suitable for collaborative filtering This module contains all the high-level functions you need in a collaborative filtering application to assemble your data, get a model and train it with a `Learner`. We will go other those in order but you can also check the [collaborative filtering tutorial](http://docs.fast.ai/tutorial.collab). Gather the data
|
#export
class TabularCollab(TabularPandas):
"Instance of `TabularPandas` suitable for collaborative filtering (with no continuous variable)"
with_cont=False
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
This is just to use the internal of the tabular application, don't worry about it.
|
#export
class CollabDataLoaders(DataLoaders):
"Base `DataLoaders` for collaborative filtering."
@delegates(DataLoaders.from_dblock)
@classmethod
def from_df(cls, ratings, valid_pct=0.2, user_name=None, item_name=None, rating_name=None, seed=None, path='.', **kwargs):
"Create a `DataLoaders` suitable for collaborative filtering from `ratings`."
user_name = ifnone(user_name, ratings.columns[0])
item_name = ifnone(item_name, ratings.columns[1])
rating_name = ifnone(rating_name, ratings.columns[2])
cat_names = [user_name,item_name]
splits = RandomSplitter(valid_pct=valid_pct, seed=seed)(range_of(ratings))
to = TabularCollab(ratings, [Categorify], cat_names, y_names=[rating_name], y_block=TransformBlock(), splits=splits)
return to.dataloaders(path=path, **kwargs)
@classmethod
def from_csv(cls, csv, **kwargs):
"Create a `DataLoaders` suitable for collaborative filtering from `csv`."
return cls.from_df(pd.read_csv(csv), **kwargs)
CollabDataLoaders.from_csv = delegates(to=CollabDataLoaders.from_df)(CollabDataLoaders.from_csv)
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
This class should not be used directly, one of the factory methods should be preferred instead. All those factory methods accept as arguments:- `valid_pct`: the random percentage of the dataset to set aside for validation (with an optional `seed`)- `user_name`: the name of the column containing the user (defaults to the first column)- `item_name`: the name of the column containing the item (defaults to the second column)- `rating_name`: the name of the column containing the rating (defaults to the third column)- `path`: the folder where to work- `bs`: the batch size- `val_bs`: the batch size for the validation `DataLoader` (defaults to `bs`)- `shuffle_train`: if we shuffle the training `DataLoader` or not- `device`: the PyTorch device to use (defaults to `default_device()`)
|
show_doc(CollabDataLoaders.from_df)
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
Let's see how this works on an example:
|
path = untar_data(URLs.ML_SAMPLE)
ratings = pd.read_csv(path/'ratings.csv')
ratings.head()
dls = CollabDataLoaders.from_df(ratings, bs=64)
dls.show_batch()
show_doc(CollabDataLoaders.from_csv)
dls = CollabDataLoaders.from_csv(path/'ratings.csv', bs=64)
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
Models fastai provides two kinds of models for collaborative filtering: a dot-product model and a neural net.
|
#export
class EmbeddingDotBias(Module):
"Base dot model for collaborative filtering."
def __init__(self, n_factors, n_users, n_items, y_range=None):
self.y_range = y_range
(self.u_weight, self.i_weight, self.u_bias, self.i_bias) = [Embedding(*o) for o in [
(n_users, n_factors), (n_items, n_factors), (n_users,1), (n_items,1)
]]
def forward(self, x):
users,items = x[:,0],x[:,1]
dot = self.u_weight(users)* self.i_weight(items)
res = dot.sum(1) + self.u_bias(users).squeeze() + self.i_bias(items).squeeze()
if self.y_range is None: return res
return torch.sigmoid(res) * (self.y_range[1]-self.y_range[0]) + self.y_range[0]
@classmethod
def from_classes(cls, n_factors, classes, user=None, item=None, y_range=None):
"Build a model with `n_factors` by inferring `n_users` and `n_items` from `classes`"
if user is None: user = list(classes.keys())[0]
if item is None: item = list(classes.keys())[1]
res = cls(n_factors, len(classes[user]), len(classes[item]), y_range=y_range)
res.classes,res.user,res.item = classes,user,item
return res
def _get_idx(self, arr, is_item=True):
"Fetch item or user (based on `is_item`) for all in `arr`"
assert hasattr(self, 'classes'), "Build your model with `EmbeddingDotBias.from_classes` to use this functionality."
classes = self.classes[self.item] if is_item else self.classes[self.user]
c2i = {v:k for k,v in enumerate(classes)}
try: return tensor([c2i[o] for o in arr])
except Exception as e:
print(f"""You're trying to access {'an item' if is_item else 'a user'} that isn't in the training data.
If it was in your original data, it may have been split such that it's only in the validation set now.""")
def bias(self, arr, is_item=True):
"Bias for item or user (based on `is_item`) for all in `arr`"
idx = self._get_idx(arr, is_item)
layer = (self.i_bias if is_item else self.u_bias).eval().cpu()
return to_detach(layer(idx).squeeze(),gather=False)
def weight(self, arr, is_item=True):
"Weight for item or user (based on `is_item`) for all in `arr`"
idx = self._get_idx(arr, is_item)
layer = (self.i_weight if is_item else self.u_weight).eval().cpu()
return to_detach(layer(idx),gather=False)
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
The model is built with `n_factors` (the length of the internal vectors), `n_users` and `n_items`. For a given user and item, it grabs the corresponding weights and bias and returns``` pythontorch.dot(user_w, item_w) + user_b + item_b```Optionally, if `y_range` is passed, it applies a `SigmoidRange` to that result.
|
x,y = dls.one_batch()
model = EmbeddingDotBias(50, len(dls.classes['userId']), len(dls.classes['movieId']), y_range=(0,5)
).to(x.device)
out = model(x)
assert (0 <= out).all() and (out <= 5).all()
show_doc(EmbeddingDotBias.from_classes)
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
`y_range` is passed to the main init. `user` and `item` are the names of the keys for users and items in `classes` (default to the first and second key respectively). `classes` is expected to be a dictionary key to list of categories like the result of `dls.classes` in a `CollabDataLoaders`:
|
dls.classes
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
Let's see how it can be used in practice:
|
model = EmbeddingDotBias.from_classes(50, dls.classes, y_range=(0,5)
).to(x.device)
out = model(x)
assert (0 <= out).all() and (out <= 5).all()
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
Two convenience methods are added to easily access the weights and bias when a model is created with `EmbeddingDotBias.from_classes`:
|
show_doc(EmbeddingDotBias.weight)
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
The elements of `arr` are expected to be class names (which is why the model needs to be created with `EmbeddingDotBias.from_classes`)
|
mov = dls.classes['movieId'][42]
w = model.weight([mov])
test_eq(w, model.i_weight(tensor([42])))
show_doc(EmbeddingDotBias.bias)
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
The elements of `arr` are expected to be class names (which is why the model needs to be created with `EmbeddingDotBias.from_classes`)
|
mov = dls.classes['movieId'][42]
b = model.bias([mov])
test_eq(b, model.i_bias(tensor([42])))
#export
class EmbeddingNN(TabularModel):
"Subclass `TabularModel` to create a NN suitable for collaborative filtering."
@delegates(TabularModel.__init__)
def __init__(self, emb_szs, layers, **kwargs):
super().__init__(emb_szs=emb_szs, n_cont=0, out_sz=1, layers=layers, **kwargs)
show_doc(EmbeddingNN)
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
`emb_szs` should be a list of two tuples, one for the users, one for the items, each tuple containing the number of users/items and the corresponding embedding size (the function `get_emb_sz` can give a good default). All the other arguments are passed to `TabularModel`.
|
emb_szs = get_emb_sz(dls.train_ds, {})
model = EmbeddingNN(emb_szs, [50], y_range=(0,5)
).to(x.device)
out = model(x)
assert (0 <= out).all() and (out <= 5).all()
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
Create a `Learner` The following function lets us quickly create a `Learner` for collaborative filtering from the data.
|
# export
@log_args(to_return=True, but_as=Learner.__init__)
@delegates(Learner.__init__)
def collab_learner(dls, n_factors=50, use_nn=False, emb_szs=None, layers=None, config=None, y_range=None, loss_func=None, **kwargs):
"Create a Learner for collaborative filtering on `dls`."
emb_szs = get_emb_sz(dls, ifnone(emb_szs, {}))
if loss_func is None: loss_func = MSELossFlat()
if config is None: config = tabular_config()
if y_range is not None: config['y_range'] = y_range
if layers is None: layers = [n_factors]
if use_nn: model = EmbeddingNN(emb_szs=emb_szs, layers=layers, **config)
else: model = EmbeddingDotBias.from_classes(n_factors, dls.classes, y_range=y_range)
return Learner(dls, model, loss_func=loss_func, **kwargs)
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
If `use_nn=False`, the model used is an `EmbeddingDotBias` with `n_factors` and `y_range`. Otherwise, it's a `EmbeddingNN` for which you can pass `emb_szs` (will be inferred from the `dls` with `get_emb_sz` if you don't provide any), `layers` (defaults to `[n_factors]`) `y_range`, and a `config` that you can create with `tabular_config` to customize your model. `loss_func` will default to `MSELossFlat` and all the other arguments are passed to `Learner`.
|
learn = collab_learner(dls, y_range=(0,5))
learn.fit_one_cycle(1)
|
_____no_output_____
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
Export -
|
#hide
from nbdev.export import *
notebook2script()
|
Converted 00_torch_core.ipynb.
Converted 01_layers.ipynb.
Converted 02_data.load.ipynb.
Converted 03_data.core.ipynb.
Converted 04_data.external.ipynb.
Converted 05_data.transforms.ipynb.
Converted 06_data.block.ipynb.
Converted 07_vision.core.ipynb.
Converted 08_vision.data.ipynb.
Converted 09_vision.augment.ipynb.
Converted 09b_vision.utils.ipynb.
Converted 09c_vision.widgets.ipynb.
Converted 10_tutorial.pets.ipynb.
Converted 11_vision.models.xresnet.ipynb.
Converted 12_optimizer.ipynb.
Converted 13_callback.core.ipynb.
Converted 13a_learner.ipynb.
Converted 13b_metrics.ipynb.
Converted 14_callback.schedule.ipynb.
Converted 14a_callback.data.ipynb.
Converted 15_callback.hook.ipynb.
Converted 15a_vision.models.unet.ipynb.
Converted 16_callback.progress.ipynb.
Converted 17_callback.tracker.ipynb.
Converted 18_callback.fp16.ipynb.
Converted 18a_callback.training.ipynb.
Converted 19_callback.mixup.ipynb.
Converted 20_interpret.ipynb.
Converted 20a_distributed.ipynb.
Converted 21_vision.learner.ipynb.
Converted 22_tutorial.imagenette.ipynb.
Converted 23_tutorial.vision.ipynb.
Converted 24_tutorial.siamese.ipynb.
Converted 24_vision.gan.ipynb.
Converted 30_text.core.ipynb.
Converted 31_text.data.ipynb.
Converted 32_text.models.awdlstm.ipynb.
Converted 33_text.models.core.ipynb.
Converted 34_callback.rnn.ipynb.
Converted 35_tutorial.wikitext.ipynb.
Converted 36_text.models.qrnn.ipynb.
Converted 37_text.learner.ipynb.
Converted 38_tutorial.text.ipynb.
Converted 39_tutorial.transformers.ipynb.
Converted 40_tabular.core.ipynb.
Converted 41_tabular.data.ipynb.
Converted 42_tabular.model.ipynb.
Converted 43_tabular.learner.ipynb.
Converted 44_tutorial.tabular.ipynb.
Converted 45_collab.ipynb.
Converted 46_tutorial.collab.ipynb.
Converted 50_tutorial.datablock.ipynb.
Converted 60_medical.imaging.ipynb.
Converted 61_tutorial.medical_imaging.ipynb.
Converted 65_medical.text.ipynb.
Converted 70_callback.wandb.ipynb.
Converted 71_callback.tensorboard.ipynb.
Converted 72_callback.neptune.ipynb.
Converted 73_callback.captum.ipynb.
Converted 74_callback.cutmix.ipynb.
Converted 97_test_utils.ipynb.
Converted 99_pytorch_doc.ipynb.
Converted index.ipynb.
Converted tutorial.ipynb.
|
Apache-2.0
|
nbs/45_collab.ipynb
|
ldanilov/fastai
|
Model
|
def get_3d_head(p=0.0):
pool, feat = (nn.AdaptiveAvgPool3d(1), 64)
m = nn.Sequential(Batchify(),
ConvLayer(512,512,stride=2,ndim=3), # 8
ConvLayer(512,1024,stride=2,ndim=3), # 4
ConvLayer(1024,1024,stride=2,ndim=3), # 2
nn.AdaptiveAvgPool3d((1, 1, 1)), Batchify(), Flat3d(), nn.Dropout(p),
nn.Linear(1024, 6))
init_cnn(m)
return m
m = get_3d_head()
config=dict(custom_head=m)
learn = get_learner(dls, xresnet18, get_loss(), config=config)
hook = ReshapeBodyHook(learn.model[0])
learn.add_cb(RowLoss())
# learn.load(f'runs/baseline_stg1_xresnet18-3', strict=False)
name = 'trainfull3d_labels_partial3d_new'
|
_____no_output_____
|
Apache-2.0
|
04_trainfull3d/04_trainfull3d_labels_01_partial3d.ipynb
|
bearpelican/rsna_retro
|
Training
|
learn.lr_find()
do_fit(learn, 8, 1e-3)
learn.save(f'runs/{name}-1')
learn.load(f'runs/{name}-1')
learn.dls = get_3d_dls_aug(Meta.df_comb, sz=256, bs=12, grps=Meta.grps_stg1)
do_fit(learn, 4, 1e-4)
learn.save(f'runs/{name}-2')
learn.load(f'runs/{name}-2')
learn.dls = get_3d_dls_aug(Meta.df_comb, sz=384, bs=4, path=path_jpg, grps=Meta.grps_stg1)
do_fit(learn, 2, 1e-5)
learn.save(f'runs/{name}-3')
|
_____no_output_____
|
Apache-2.0
|
04_trainfull3d/04_trainfull3d_labels_01_partial3d.ipynb
|
bearpelican/rsna_retro
|
Import Modules
|
import cv2
import numpy as np
from google.colab.patches import cv2_imshow
|
_____no_output_____
|
MIT
|
ImageResize/Image_Scaling/Image_scaling.ipynb
|
noviicee/Image-Processing-OpenCV
|
Load Image
|
#image is loaded using cv2.imread() method,here flag is 0 ,specifies to load image in GRAYSCALE mode.
'''
Syntax:
cv2.imread(path,flag)
Parameters:
path: string representing the path of the image to be read.
flag: specifies the way in which image should be read.
'''
img=cv2.imread("input.png",0)
cv2_imshow(img)
|
_____no_output_____
|
MIT
|
ImageResize/Image_Scaling/Image_scaling.ipynb
|
noviicee/Image-Processing-OpenCV
|
Apply scaling Operation
|
# To perform scaling operation,cv2.resize() method is used.
'''
Syntax:
cv2.resize(image,(width,height)=None,fx=1,fy=1,interpolation)
Parameters:
image: input image.
(width,height): determining the size of output image ; optional parameter.
fx: scaling factor for x-axis,default=1.
fy: scaling factor for y-axis,default=1.
interpolation: interpolation method to be used.
'''
scaled_up_x=cv2.resize(img,None,fx=2,fy=1,interpolation=cv2.INTER_CUBIC)
scaled_down_x=cv2.resize(img,None,fx=0.5,fy=1,interpolation=cv2.INTER_LINEAR)
scaled_up_y=cv2.resize(img,None,fx=1,fy=2,interpolation=cv2.INTER_CUBIC)
scaled_down_y=cv2.resize(img,None,fx=1,fy=0.5,interpolation=cv2.INTER_LINEAR)
|
_____no_output_____
|
MIT
|
ImageResize/Image_Scaling/Image_scaling.ipynb
|
noviicee/Image-Processing-OpenCV
|
Display the scaled image
|
cv2_imshow(scaled_up_x)
cv2_imshow(scaled_down_x)
cv2_imshow(scaled_up_y)
cv2_imshow(scaled_down_y)
|
_____no_output_____
|
MIT
|
ImageResize/Image_Scaling/Image_scaling.ipynb
|
noviicee/Image-Processing-OpenCV
|
***Introduction to Radar Using Python and MATLAB*** Andy Harrison - Copyright (C) 2019 Artech House Pulse Train Ambiguity Function*** Referring to Section 8.6.1, the amibguity function for a coherent pulse train is found by employing the generic waveform technique outlined in Section 8.6.3.*** Begin by getting the library path
|
import lib_path
|
_____no_output_____
|
Apache-2.0
|
jupyter/Chapter08/pulse_train_ambiguity.ipynb
|
mberkanbicer/software
|
Set the pulsewidth (s), the pulse repetition interval (s) and the number of pulses
|
pulsewidth = 0.4
pri = 1.0
number_of_pulses = 6
|
_____no_output_____
|
Apache-2.0
|
jupyter/Chapter08/pulse_train_ambiguity.ipynb
|
mberkanbicer/software
|
Generate the time delay (s) using the `linspace` routine from `scipy`
|
from numpy import linspace
# Set the time delay
time_delay = linspace(-number_of_pulses * pri, number_of_pulses * pri, 5000)
|
_____no_output_____
|
Apache-2.0
|
jupyter/Chapter08/pulse_train_ambiguity.ipynb
|
mberkanbicer/software
|
Calculate the ambiguity function for the pulse train
|
from Libs.ambiguity.ambiguity_function import pulse_train
from numpy import finfo
ambiguity = pulse_train(time_delay, finfo(float).eps, pulsewidth, pri, number_of_pulses)
|
_____no_output_____
|
Apache-2.0
|
jupyter/Chapter08/pulse_train_ambiguity.ipynb
|
mberkanbicer/software
|
Plot the zero-Doppler cut using the `matplotlib` routines
|
from matplotlib import pyplot as plt
# Set the figure size
plt.rcParams["figure.figsize"] = (15, 10)
# Plot the ambiguity function
plt.plot(time_delay, ambiguity, '')
# Set the x and y axis labels
plt.xlabel("Time (s)", size=12)
plt.ylabel("Relative Amplitude", size=12)
# Turn on the grid
plt.grid(linestyle=':', linewidth=0.5)
# Set the plot title and labels
plt.title('Pulse Train Ambiguity Function', size=14)
# Set the tick label size
plt.tick_params(labelsize=12)
|
_____no_output_____
|
Apache-2.0
|
jupyter/Chapter08/pulse_train_ambiguity.ipynb
|
mberkanbicer/software
|
Set the Doppler mismatch frequencies using the `linspace` routine
|
doppler_frequency = linspace(-2.0 / pulsewidth, 2.0 / pulsewidth, 1000)
|
_____no_output_____
|
Apache-2.0
|
jupyter/Chapter08/pulse_train_ambiguity.ipynb
|
mberkanbicer/software
|
Calculate the ambiguity function for the pulse train
|
ambiguity = pulse_train(finfo(float).eps, doppler_frequency, pulsewidth, pri, number_of_pulses)
|
_____no_output_____
|
Apache-2.0
|
jupyter/Chapter08/pulse_train_ambiguity.ipynb
|
mberkanbicer/software
|
Display the zero-range cut for the pulse train
|
plt.plot(doppler_frequency, ambiguity, '')
# Set the x and y axis labels
plt.xlabel("Doppler (Hz)", size=12)
plt.ylabel("Relative Amplitude", size=12)
# Turn on the grid
plt.grid(linestyle=':', linewidth=0.5)
# Set the plot title and labels
plt.title('Pulse Train Ambiguity Function', size=14)
# Set the tick label size
plt.tick_params(labelsize=12)
|
_____no_output_____
|
Apache-2.0
|
jupyter/Chapter08/pulse_train_ambiguity.ipynb
|
mberkanbicer/software
|
Set the time delay and Doppler mismatch frequency and create the two-dimensional grid using the `meshgrid` routine from `scipy`
|
from numpy import meshgrid
# Set the time delay
time_delay = linspace(-number_of_pulses * pri, number_of_pulses * pri, 1000)
# Set the Doppler mismatch
doppler_frequency = linspace(-2.0 / pulsewidth, 2.0 / pulsewidth, 1000)
# Create the grid
t, f = meshgrid(time_delay, doppler_frequency)
|
_____no_output_____
|
Apache-2.0
|
jupyter/Chapter08/pulse_train_ambiguity.ipynb
|
mberkanbicer/software
|
Calculate the ambiguity function for the pulse train
|
ambiguity = pulse_train(t, f, pulsewidth, pri, number_of_pulses)
|
_____no_output_____
|
Apache-2.0
|
jupyter/Chapter08/pulse_train_ambiguity.ipynb
|
mberkanbicer/software
|
Display the two-dimensional contour plot for the pulse train ambiguity function
|
# Plot the ambiguity function
from numpy import finfo
plt.contour(t, f, ambiguity + finfo('float').eps, 20, cmap='jet', vmin=-0.2, vmax=1.0)
# Set the x and y axis labels
plt.xlabel("Time (s)", size=12)
plt.ylabel("Doppler (Hz)", size=12)
# Turn on the grid
plt.grid(linestyle=':', linewidth=0.5)
# Set the plot title and labels
plt.title('Pulse Pulse Ambiguity Function', size=14)
# Set the tick label size
plt.tick_params(labelsize=12)
|
_____no_output_____
|
Apache-2.0
|
jupyter/Chapter08/pulse_train_ambiguity.ipynb
|
mberkanbicer/software
|
============================================4D Neuroimaging/BTi phantom dataset tutorial============================================Here we read 4DBTi epochs data obtained with a spherical phantomusing four different dipole locations. For each condition wecompute evoked data and compute dipole fits.Data are provided by Jean-Michel Badier from MEG center in Marseille, France.
|
# Authors: Alex Gramfort <[email protected]>
#
# License: BSD (3-clause)
import os.path as op
import numpy as np
from mayavi import mlab
from mne.datasets import phantom_4dbti
import mne
|
_____no_output_____
|
BSD-3-Clause
|
stable/_downloads/a68c968ba9eafa2b1315cbf9e139eee3/plot_phantom_4DBTi.ipynb
|
drammock/mne-tools.github.io
|
Read data and compute a dipole fit at the peak of the evoked response
|
data_path = phantom_4dbti.data_path()
raw_fname = op.join(data_path, '%d/e,rfhp1.0Hz')
dipoles = list()
sphere = mne.make_sphere_model(r0=(0., 0., 0.), head_radius=0.080)
t0 = 0.07 # peak of the response
pos = np.empty((4, 3))
for ii in range(4):
raw = mne.io.read_raw_bti(raw_fname % (ii + 1,),
rename_channels=False, preload=True)
raw.info['bads'] = ['A173', 'A213', 'A232']
events = mne.find_events(raw, 'TRIGGER', mask=4350, mask_type='not_and')
epochs = mne.Epochs(raw, events=events, event_id=8192, tmin=-0.2, tmax=0.4,
preload=True)
evoked = epochs.average()
evoked.plot(time_unit='s')
cov = mne.compute_covariance(epochs, tmax=0.)
dip = mne.fit_dipole(evoked.copy().crop(t0, t0), cov, sphere)[0]
pos[ii] = dip.pos[0]
|
_____no_output_____
|
BSD-3-Clause
|
stable/_downloads/a68c968ba9eafa2b1315cbf9e139eee3/plot_phantom_4DBTi.ipynb
|
drammock/mne-tools.github.io
|
Compute localisation errors
|
actual_pos = 0.01 * np.array([[0.16, 1.61, 5.13],
[0.17, 1.35, 4.15],
[0.16, 1.05, 3.19],
[0.13, 0.80, 2.26]])
actual_pos = np.dot(actual_pos, [[0, 1, 0], [-1, 0, 0], [0, 0, 1]])
errors = 1e3 * np.linalg.norm(actual_pos - pos, axis=1)
print("errors (mm) : %s" % errors)
|
_____no_output_____
|
BSD-3-Clause
|
stable/_downloads/a68c968ba9eafa2b1315cbf9e139eee3/plot_phantom_4DBTi.ipynb
|
drammock/mne-tools.github.io
|
Plot the dipoles in 3D
|
def plot_pos(pos, color=(0., 0., 0.)):
mlab.points3d(pos[:, 0], pos[:, 1], pos[:, 2], scale_factor=0.005,
color=color)
mne.viz.plot_alignment(evoked.info, bem=sphere, surfaces=[])
# Plot the position of the actual dipole
plot_pos(actual_pos, color=(1., 0., 0.))
# Plot the position of the estimated dipole
plot_pos(pos, color=(1., 1., 0.))
|
_____no_output_____
|
BSD-3-Clause
|
stable/_downloads/a68c968ba9eafa2b1315cbf9e139eee3/plot_phantom_4DBTi.ipynb
|
drammock/mne-tools.github.io
|
F1, Precision Recall, and Confusion Matrix
|
from sklearn.metrics import precision_recall_fscore_support
from sklearn.metrics import recall_score
from sklearn.metrics import classification_report
y_prediction = model.predict_classes(X_test)
y_prediction.reshape(-1,1)
print("Recall score:"+ str(recall_score(y_test, y_prediction)))
print(classification_report(y_test, y_prediction,
target_names=["default", "non_default"]))
import itertools
import matplotlib.pyplot as plt
from sklearn.metrics import confusion_matrix
def plot_confusion_matrix(cm, classes,
normalize=False,
title='Confusion matrix',
cmap=plt.cm.Blues):
"""
This function prints and plots the confusion matrix.
Normalization can be applied by setting `normalize=True`.
"""
if normalize:
cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]
print("Normalized confusion matrix")
else:
print('Confusion matrix, without normalization')
print(cm)
plt.imshow(cm, interpolation='nearest', cmap=cmap)
plt.title(title)
plt.colorbar()
tick_marks = np.arange(len(classes))
plt.xticks(tick_marks, classes, rotation=45)
plt.yticks(tick_marks, classes)
fmt = '.2f' if normalize else 'd'
thresh = cm.max() / 2.
for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):
plt.text(j, i, format(cm[i, j], fmt),
horizontalalignment="center",
color="red" if cm[i, j] > thresh else "black")
plt.tight_layout()
plt.ylabel('True label')
plt.xlabel('Predicted label')
# Compute confusion matrix
cnf_matrix = confusion_matrix(y_test, y_prediction)
np.set_printoptions(precision=2)
# Plot non-normalized confusion matrix
plt.figure()
plot_confusion_matrix(cnf_matrix, classes=['Defualt', 'Non_default'],
title='Confusion matrix, without normalization')
# Plot normalized confusion matrix
plt.figure()
plot_confusion_matrix(cnf_matrix, classes=['Defualt', 'Non_default'], normalize=True,
title='Normalized confusion matrix')
plt.show()
|
Confusion matrix, without normalization
[[4687 0]
[1313 0]]
Normalized confusion matrix
[[1. 0.]
[1. 0.]]
|
MIT
|
Model/3-NeuralNetwork4.ipynb
|
skawns0724/KOSA-Big-Data_Vision
|
Nonlinear recharge models*R.A. Collenteur, University of Graz*This notebook explains the use of the `RechargeModel` stress model to simulate the combined effect of precipitation and potential evaporation on the groundwater levels. For the computation of the groundwater recharge, three recharge models are currently available:- `Linear` ([Berendrecht et al., 2003](References); [von Asmuth et al., 2008](References))- `Berendrecht` ([Berendrecht et al., 2006](References))- `FlexModel` ([Collenteur et al., 2021](References))The first model is a simple linear function of precipitation and potential evaporation while the latter two are simulate a nonlinear response of recharge to precipitation using a soil-water balance concepts. Detailed descriptions of these models can be found in articles listed in the [References](References) at the end of this notebook. Tip To run this notebook and the related non-linear recharge models, it is strongly recommended to install Numba (http://numba.pydata.org). This Just-In-Time (JIT) compiler compiles the computationally intensive part of the recharge calculation, making the non-linear model as fast as the Linear recharge model.
|
import pandas as pd
import pastas as ps
import matplotlib.pyplot as plt
ps.show_versions(numba=True)
ps.set_log_level("INFO")
|
Python version: 3.8.2 (default, Mar 25 2020, 11:22:43)
[Clang 4.0.1 (tags/RELEASE_401/final)]
Numpy version: 1.20.2
Scipy version: 1.6.2
Pandas version: 1.1.5
Pastas version: 0.18.0b
Matplotlib version: 3.3.4
numba version: 0.51.2
|
MIT
|
examples/notebooks/07_non_linear_recharge.ipynb
|
pastas/pastas
|
Read Input dataInput data handling is similar to other stressmodels. The only thing that is necessary to check is that the precipitation and evaporation are provided in mm/day. This is necessary because the parameters for the nonlinear recharge models are defined in mm for the length unit and days for the time unit. It is possible to use other units, but this would require manually setting the initial values and parameter boundaries for the recharge models.
|
head = pd.read_csv("../data/B32C0639001.csv", parse_dates=['date'],
index_col='date', squeeze=True)
# Make this millimeters per day
evap = ps.read_knmi("../data/etmgeg_260.txt", variables="EV24").series * 1e3
rain = ps.read_knmi("../data/etmgeg_260.txt", variables="RH").series * 1e3
fig, axes = plt.subplots(3,1, figsize=(10,6), sharex=True)
head.plot(ax=axes[0], x_compat=True, linestyle=" ", marker=".")
evap.plot(ax=axes[1], x_compat=True)
rain.plot(ax=axes[2], x_compat=True)
axes[0].set_ylabel("Head [m]")
axes[1].set_ylabel("Evap [mm/d]")
axes[2].set_ylabel("Rain [mm/d]")
plt.xlim("1985", "2005");
|
INFO: Inferred frequency for time series EV24 260: freq=D
INFO: Inferred frequency for time series RH 260: freq=D
|
MIT
|
examples/notebooks/07_non_linear_recharge.ipynb
|
pastas/pastas
|
Make a basic modelThe normal workflow may be used to create and calibrate the model.1. Create a Pastas `Model` instance2. Choose a recharge model. All recharge models can be accessed through the recharge subpackage (`ps.rch`).3. Create a `RechargeModel` object and add it to the model4. Solve and visualize the model
|
ml = ps.Model(head)
# Select a recharge model
rch = ps.rch.FlexModel()
#rch = ps.rch.Berendrecht()
#rch = ps.rch.Linear()
rm = ps.RechargeModel(rain, evap, recharge=rch, rfunc=ps.Gamma, name="rch")
ml.add_stressmodel(rm)
ml.solve(noise=True, tmin="1990", report="basic")
ml.plots.results(figsize=(10,6));
|
INFO: Cannot determine frequency of series head: freq=None. The time series is irregular.
INFO: Inferred frequency for time series RH 260: freq=D
INFO: Inferred frequency for time series EV24 260: freq=D
|
MIT
|
examples/notebooks/07_non_linear_recharge.ipynb
|
pastas/pastas
|
Analyze the estimated recharge fluxAfter the parameter estimation we can take a look at the recharge flux computed by the model. The flux is easy to obtain using the `get_stress` method of the model object, which automatically provides the optimal parameter values that were just estimated. After this, we can for example look at the yearly recharge flux estimated by the Pastas model.
|
recharge = ml.get_stress("rch").resample("A").sum()
ax = recharge.plot.bar(figsize=(10,3))
ax.set_xticklabels(recharge.index.year)
plt.ylabel("Recharge [mm/year]");
|
_____no_output_____
|
MIT
|
examples/notebooks/07_non_linear_recharge.ipynb
|
pastas/pastas
|
Place Stock Trades into Senator Dataframe 1. Understand the Senator Trading Report (STR) Dataframe
|
import pandas as pd
#https://docs.google.com/spreadsheets/d/1lH_LpTgRlfzKvpRnWYgoxlkWvJj0v1r3zN3CeWMAgqI/edit?usp=sharing
try:
sen_df = pd.read_csv("Senator Stock Trades/Senate Stock Watcher 04_16_2020 All Transactions.csv")
except:
sen_df = pd.read_csv("https://github.com/pkm29/big_data_final_project/raw/master/Senate%20Stock%20Trades/Senate%20Stock%20Watcher%2004_16_2020%20All%20Transactions.csv")
sen_df.head()
sen_df.type.unique()
|
_____no_output_____
|
MIT
|
Stocks/Place Stock Trades into Senator Dataframe Ankur Edit.ipynb
|
paulmtree/Suspicious-Senator-Trading
|
There are 4 types of trades.Exchanges: Exchange 1 stock for anotherSale (Full): Selling all of their stockPurchase: Buying a stockSale (Partial): Selling some of that particular stock
|
n_exchanges = len(sen_df.loc[sen_df['type'] == "Exchange"])
n_trades = len(sen_df)
print("There are " +str(n_exchanges) +" exchange trades out of a total of " +str(n_trades)+ " trades.")
sen_df = sen_df.loc[sen_df['type'] != "Exchange"]
|
There are 84 exchange trades out of a total of 8600 trades.
|
MIT
|
Stocks/Place Stock Trades into Senator Dataframe Ankur Edit.ipynb
|
paulmtree/Suspicious-Senator-Trading
|
At this point in time, I will exclude exchange trades because they are so few and wish to build the basic structure of the project. As you can see, this would require splitting up the exchange into two rows with each company and so on. I may include this step later if time permits. There should now be 8516 trades remaining in the dataframe. Let's make sure this is so.
|
n_trades = len(sen_df)
print("There are " +str(n_trades)+ " trades in the dataframe")
n_blank_ticker = len(sen_df.loc[sen_df['ticker'] == "--"])
print("There are " +str(n_blank_ticker) +" trades w/o a ticker out of a total of " +str(n_trades)+ " trades")
sen_df = sen_df.loc[sen_df['ticker'] != "--"]
|
There are 1872 trades w/o a ticker out of a total of 8516 trades
|
MIT
|
Stocks/Place Stock Trades into Senator Dataframe Ankur Edit.ipynb
|
paulmtree/Suspicious-Senator-Trading
|
For the same reasons we excluded exchange trades, we will also exclude trades without a ticker (which all public stocks have - the ticker is their identifier on the stock exchange). Eliminating trades without a ticker takes out trades of other types of securities (corporate bonds, municipal securities, non-public stock). There should now be 6644 trades remaining in the dataframe. Let's make sure this is so.
|
n_trades = len(sen_df)
print("There are " +str(n_trades)+ " trades in the dataframe")
|
There are 6644 trades in the dataframe
|
MIT
|
Stocks/Place Stock Trades into Senator Dataframe Ankur Edit.ipynb
|
paulmtree/Suspicious-Senator-Trading
|
2. Add Data to STR Dataframe Import Data In this step we will be using company information such as market cap and industry from online lists provided by the NYSE, NASDAQ, and ASXL exchange. Links can be found here:https://stackoverflow.com/questions/25338608/download-all-stock-symbol-list-of-a-market
|
ticker_list = list()
try:
NYSE_df = pd.read_csv("NYSEcompanylist.csv")
except:
NYSE_df = pd.read_csv("https://github.com/pkm29/big_data_final_project/raw/master/Stocks/NYSEcompanylist.csv")
try:
NASDAQ_df = pd.read_csv("NASDAQcompanylist.csv")
except:
NASDAQ_df = pd.read_csv("https://github.com/pkm29/big_data_final_project/raw/master/Stocks/NASDAQcompanylist.csv")
ticker_list.append(NYSE_df)
ticker_list.append(NASDAQ_df)
NYSE_df.head()
NASDAQ_df.head()
"""
Add data for Berkshire Hathaway, Lions Gate Entertainment, and Royal Dutch Shell to the NYSE company list. While
#these companies are in the company list, their fields are empty. Also, change the tickers of these companies to
#match Senate Stock Data (since dashes are used instead of periods in that dataset, we make sure the same is true
in the NYSE company list). What matters is consistent convention here.
"""
row_count = 0
replacement_count = 0
for row_tuple in NYSE_df.itertuples():
if replacement_count == 4:
break
if row_tuple.Symbol == "BRK.B":
#row_tuple.Symbol = "BRK-B"
NYSE_df.at[row_count, 'Symbol'] = "BRK-B"
#Shares outstanding reported in Q1 2020 financial reports, stock price from May 6, when this data is dated
#row_tuple.MarketCap = "$420.02B"
NYSE_df.at[row_count, 'MarketCap'] = "$420.02B"
#row_tuple.Sector = "Miscellaneous"
NYSE_df.at[row_count, 'Sector'] = "Miscellaneous"
#row_tuple.industry = "Conglomerate"
NYSE_df.at[row_count, 'industry'] = "Conglomerate"
replacement_count = replacement_count + 1
if row_tuple.Symbol == "LGF.B":
#row_tuple.Symbol = "LGF-B"
#Shares outstanding reported in Q1 2020 financial reports, stock price from May 6, when this data is dated
#row_tuple.MarketCap = "$14.62B"
#row_tuple.Sector = "Consumer Services"
#row_tuple.industry = "Movies/Entertainment"
NYSE_df.at[row_count, 'Symbol'] = "LGF-B"
NYSE_df.at[row_count, 'MarketCap'] = "$14.62B"
NYSE_df.at[row_count, 'Sector'] = "Consumer Services"
NYSE_df.at[row_count, 'industry'] = "Movies/Entertainment"
replacement_count = replacement_count + 1
if row_tuple.Symbol == "RDS.A":
#row_tuple.Symbol = "RDS-A"
#Shares outstanding reported in Q1 2020 financial reports, stock price from May 6, when this data is dated
#row_tuple.MarketCap = "$122.28B"
#row_tuple.Sector = "Energy"
#row_tuple.industry = "Oil & Gas Production"
NYSE_df.at[row_count, 'Symbol'] = "RDS-A"
NYSE_df.at[row_count, 'MarketCap'] = "$122.28B"
NYSE_df.at[row_count, 'Sector'] = "Energy"
NYSE_df.at[row_count, 'industry'] = "Oil & Gas Production"
replacement_count = replacement_count + 1
if row_tuple.Symbol == "RDS.B":
#row_tuple.Symbol = "RDS-B"
#Shares outstanding reported in Q1 2020 financial reports, stock price from May 6, when this data is dated
#row_tuple.MarketCap = "$122.09B"
#row_tuple.Sector = "Energy"
#row_tuple.industry = "Oil & Gas Production"
NYSE_df.at[row_count, 'Symbol'] = "RDS-B"
NYSE_df.at[row_count, 'MarketCap'] = "$122.09B"
NYSE_df.at[row_count, 'Sector'] = "Energy"
NYSE_df.at[row_count, 'industry'] = "Oil & Gas Production"
replacement_count = replacement_count + 1
row_count = row_count + 1
#Confirm changes have been made successfully
for row_tuple in NYSE_df.itertuples():
if row_tuple.Symbol == "BRK-B":
print (row_tuple)
if row_tuple.Symbol == "LGF-B":
print (row_tuple)
if row_tuple.Symbol == "RDS-A":
print (row_tuple)
if row_tuple.Symbol == "RDS-B":
print (row_tuple)
#There are also 2 instances where a wrong ticker for Berkshire Hathaway is found in the Senate Stock data
#(BRKB is used as opposed to BRK-B). Thus, we correct for those instances here.
#Find indices of these two trades
for row_tuple in sen_df.itertuples():
if row_tuple.ticker == "BRKB":
print (row_tuple)
#We can see that the indices are 1207 and 4611, so we will manually modify the ticker field of these trades.
sen_df.at[1207, 'ticker'] = "BRK-B"
sen_df.at[4611, 'ticker'] = "BRK-B"
len(sen_df)
#Get sector data for each stock trade
sector_data = list()
for row_tuple in sen_df.itertuples():
tic = row_tuple.ticker
count = 0
for row_tuple_tic in NYSE_df.itertuples():
sym = row_tuple_tic.Symbol
if tic == sym:
count = count+1
if row_tuple_tic.Sector == "n/a":
sector_data.append("none")
else:
sector_data.append(row_tuple_tic.Sector)
break
if count == 0:
for row_tuple_tic in NASDAQ_df.itertuples():
sym = row_tuple_tic.Symbol
if tic == sym:
count = count+1
if row_tuple_tic.Sector == "n/a":
sector_data.append("none")
else:
sector_data.append(row_tuple_tic.Sector)
break
if count == 0:
sector_data.append("none")
print(sector_data[0:9])
#make sure length matches number of rows in df
print(len(sector_data))
#counter for how many times the stock traded by senator not found in exchange data set
no_ticker_cnt = 0
for i in sector_data:
if i == "none":
no_ticker_cnt = no_ticker_cnt + 1
print(no_ticker_cnt)
#Get industry data for each stock trade
industry_data = list()
for row_tuple in sen_df.itertuples():
tic = row_tuple.ticker
count = 0
for row_tuple_tic in NYSE_df.itertuples():
sym = row_tuple_tic.Symbol
if tic == sym:
count = count+1
if row_tuple_tic.industry == "n/a":
industry_data.append("none")
else:
industry_data.append(row_tuple_tic.industry)
break
if count == 0:
for row_tuple_tic in NASDAQ_df.itertuples():
sym = row_tuple_tic.Symbol
if tic == sym:
count = count+1
if row_tuple_tic.industry == "n/a":
industry_data.append("none")
else:
industry_data.append(row_tuple_tic.industry)
break
if count == 0:
industry_data.append("none")
print(industry_data[0:9])
#make sure length matches number of rows in df
print(len(industry_data))
#counter for how many times the stock traded by senator not found in exchange data set
no_ticker_cnt = 0
for i in industry_data:
if i == "none":
no_ticker_cnt = no_ticker_cnt + 1
print(no_ticker_cnt)
#Get market cap data for each stock trade
mktcap_data = list()
for row_tuple in sen_df.itertuples():
tic = row_tuple.ticker
count = 0
for row_tuple_tic in NYSE_df.itertuples():
sym = row_tuple_tic.Symbol
if tic == sym:
count = count+1
if row_tuple_tic.MarketCap == "n/a":
mktcap_data.append("none")
else:
mktcap_data.append(row_tuple_tic.MarketCap)
break
if count == 0:
for row_tuple_tic in NASDAQ_df.itertuples():
sym = row_tuple_tic.Symbol
if tic == sym:
count = count+1
if row_tuple_tic.MarketCap == "n/a":
mktcap_data.append("none")
else:
mktcap_data.append(row_tuple_tic.MarketCap)
break
if count == 0:
mktcap_data.append("none")
print(mktcap_data[0:9])
#make sure length matches number of rows in df
print(len(mktcap_data))
#counter for how many times the stock traded by senator not found in exchange data set
no_ticker_cnt = 0
for i in mktcap_data:
if i == "none":
no_ticker_cnt = no_ticker_cnt + 1
print(no_ticker_cnt)
#add new columns to df
sen_df['mkt_cap'] = mktcap_data
sen_df['sector'] = sector_data
sen_df['industry'] = industry_data
sen_df = sen_df.fillna("none")
sen_df.head()
"""
Print out names of companies with missing data to find out why we have so many misses (~17% of our data).
There seem to be 3 reasons for this:
1. Companies merging with another or being acquired (or even acquiring and taking the acquired company's name - very rare)
2. Foreign companies (listed abroad)
3. American companies listed abroad - this applies to a very small number of trades
"""
from collections import Counter
company_missing_data = list()
for row_tuple in sen_df.itertuples():
if row_tuple.mkt_cap == "none":
company_missing_data.append(row_tuple.asset_description)
print(Counter(company_missing_data))
#Get a view of how many industries are found in our senate stock data.
industry_dict = Counter(industry_data)
industry_list = list()
for x in industry_dict:
industry_list.append(x)
print(industry_list[0:9])
n_industries = len(industry_list)
#since 'none' is included in our list
n_industries = n_industries - 1
print("There are " + str(n_industries) + " industries covered by the trades of senators.")
import string
industry_size_data = list()
for row_tuple in sen_df.itertuples():
industry_size = row_tuple.industry
if industry_size == 'none':
industry_size_data.append("none")
continue
size = row_tuple.mkt_cap
factor = 0
x = size.find("M")
if x != -1:
factor = 1000000
else:
factor = 1000000000
size = size.lstrip("$")
size = size.rstrip("MB")
size = float(size)
size = size*factor
if size < 500000000:
industry_size = industry_size + "1"
industry_size_data.append(industry_size)
continue
elif size < 1000000000:
industry_size = industry_size + "2"
industry_size_data.append(industry_size)
continue
elif size < 10000000000:
industry_size = industry_size + "3"
industry_size_data.append(industry_size)
continue
elif size < 50000000000:
industry_size = industry_size + "4"
industry_size_data.append(industry_size)
continue
elif size < 100000000000:
industry_size = industry_size + "5"
industry_size_data.append(industry_size)
continue
elif size < 500000000000:
industry_size = industry_size + "6"
industry_size_data.append(industry_size)
continue
else:
industry_size = industry_size + "7"
industry_size_data.append(industry_size)
continue
print(industry_size_data[0:9])
print(len(industry_size_data))
#add the new column to df
sen_df['classification'] = industry_size_data
sen_df.head()
#create a list of all the classifications per industry across whole dataframe, to get a view of the breakdown in
#classifications across each industry
classification_industry_breakdown = list()
for x in industry_list:
y = list()
for row_tuple in sen_df.itertuples():
if row_tuple.industry == x:
y.append(row_tuple.classification)
classification_industry_breakdown.append(y)
print(classification_industry_breakdown[0:9])
|
[['Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments3', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments3', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments3', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4', 'Biotechnology: Laboratory Analytical Instruments4'], ['Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components5', 'Industrial Machinery/Components5', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components5', 'Industrial Machinery/Components5', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components4', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components5', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components4', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components4', 'Industrial Machinery/Components5', 'Industrial Machinery/Components5', 'Industrial Machinery/Components5', 'Industrial Machinery/Components4', 'Industrial Machinery/Components5', 'Industrial Machinery/Components6', 'Industrial Machinery/Components5', 'Industrial Machinery/Components6', 'Industrial Machinery/Components4', 'Industrial Machinery/Components3', 'Industrial Machinery/Components6', 'Industrial Machinery/Components5', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components5', 'Industrial Machinery/Components5', 'Industrial Machinery/Components4', 'Industrial Machinery/Components5', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components5', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components4', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components5', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components5', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components2', 'Industrial Machinery/Components4', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components4', 'Industrial Machinery/Components3', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components3', 'Industrial Machinery/Components4', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6', 'Industrial Machinery/Components6', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components4', 'Industrial Machinery/Components2', 'Industrial Machinery/Components4', 'Industrial Machinery/Components3', 'Industrial Machinery/Components3', 'Industrial Machinery/Components4', 'Industrial Machinery/Components3', 'Industrial Machinery/Components2', 'Industrial Machinery/Components4', 'Industrial Machinery/Components6'], ['none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none', 'none'], ['Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings4', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings3', 'Paints/Coatings4', 'Paints/Coatings4', 'Paints/Coatings4', 'Paints/Coatings3'], ['Building operators4', 'Building operators4', 'Building operators4', 'Building operators4', 'Building operators4', 'Building operators3'], ['Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks3', 'Major Banks3', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks3', 'Major Banks6', 'Major Banks3', 'Major Banks4', 'Major Banks5', 'Major Banks4', 'Major Banks3', 'Major Banks3', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks4', 'Major Banks6', 'Major Banks6', 'Major Banks4', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks6', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks6', 'Major Banks3', 'Major Banks5', 'Major Banks6', 'Major Banks3', 'Major Banks5', 'Major Banks5', 'Major Banks6', 'Major Banks4', 'Major Banks6', 'Major Banks4', 'Major Banks6', 'Major Banks5', 'Major Banks4', 'Major Banks3', 'Major Banks4', 'Major Banks3', 'Major Banks3', 'Major Banks6', 'Major Banks6', 'Major Banks3', 'Major Banks6', 'Major Banks3', 'Major Banks6', 'Major Banks6', 'Major Banks3', 'Major Banks3', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks3', 'Major Banks3', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks4', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks6', 'Major Banks3', 'Major Banks5', 'Major Banks6', 'Major Banks3', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks6', 'Major Banks4', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks5', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks3', 'Major Banks6', 'Major Banks3', 'Major Banks3', 'Major Banks6', 'Major Banks3', 'Major Banks3', 'Major Banks4', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks6', 'Major Banks4', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks4', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks4', 'Major Banks3', 'Major Banks5', 'Major Banks4', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks6', 'Major Banks5', 'Major Banks4', 'Major Banks4', 'Major Banks4', 'Major Banks6', 'Major Banks3', 'Major Banks6', 'Major Banks5', 'Major Banks6', 'Major Banks5', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks4', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks4', 'Major Banks4', 'Major Banks4', 'Major Banks6', 'Major Banks3', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks3', 'Major Banks4', 'Major Banks4', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks4', 'Major Banks5', 'Major Banks3', 'Major Banks4', 'Major Banks4', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks4', 'Major Banks4', 'Major Banks4', 'Major Banks3', 'Major Banks4', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks6', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks4', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks4', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks4', 'Major Banks4', 'Major Banks4', 'Major Banks3', 'Major Banks4', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks4', 'Major Banks6', 'Major Banks4', 'Major Banks2', 'Major Banks5', 'Major Banks5', 'Major Banks5', 'Major Banks6', 'Major Banks3', 'Major Banks4', 'Major Banks4', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks4', 'Major Banks6', 'Major Banks6', 'Major Banks3', 'Major Banks3', 'Major Banks5', 'Major Banks5', 'Major Banks5', 'Major Banks4', 'Major Banks6', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks3', 'Major Banks6', 'Major Banks4', 'Major Banks6', 'Major Banks2', 'Major Banks5', 'Major Banks6', 'Major Banks3', 'Major Banks3', 'Major Banks2', 'Major Banks3', 'Major Banks6', 'Major Banks4', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks5', 'Major Banks6', 'Major Banks6', 'Major Banks6', 'Major Banks5', 'Major Banks5'], ['Semiconductors5', 'Semiconductors5', 'Semiconductors5', 'Semiconductors5', 'Semiconductors5', 'Semiconductors4', 'Semiconductors6', 'Semiconductors6', 'Semiconductors3', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors4', 'Semiconductors4', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors4', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors4', 'Semiconductors6', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors5', 'Semiconductors4', 'Semiconductors6', 'Semiconductors6', 'Semiconductors5', 'Semiconductors6', 'Semiconductors5', 'Semiconductors3', 'Semiconductors5', 'Semiconductors6', 'Semiconductors6', 'Semiconductors4', 'Semiconductors4', 'Semiconductors3', 'Semiconductors6', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors6', 'Semiconductors4', 'Semiconductors6', 'Semiconductors4', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors3', 'Semiconductors3', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors3', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors6', 'Semiconductors4', 'Semiconductors4', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors6', 'Semiconductors3', 'Semiconductors4', 'Semiconductors3', 'Semiconductors3', 'Semiconductors4', 'Semiconductors3', 'Semiconductors3', 'Semiconductors3', 'Semiconductors4', 'Semiconductors5', 'Semiconductors3', 'Semiconductors3', 'Semiconductors3', 'Semiconductors6', 'Semiconductors6', 'Semiconductors3', 'Semiconductors3', 'Semiconductors3', 'Semiconductors3', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors4', 'Semiconductors3', 'Semiconductors4'], ['Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals1', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals5', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals5', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals5', 'Major Pharmaceuticals1', 'Major Pharmaceuticals6', 'Major Pharmaceuticals5', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals4', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals2', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals5', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals4', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals4', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals5', 'Major Pharmaceuticals5', 'Major Pharmaceuticals6', 'Major Pharmaceuticals5', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals4', 'Major Pharmaceuticals5', 'Major Pharmaceuticals4', 'Major Pharmaceuticals5', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals4', 'Major Pharmaceuticals6', 'Major Pharmaceuticals5', 'Major Pharmaceuticals5', 'Major Pharmaceuticals5', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals5', 'Major Pharmaceuticals5', 'Major Pharmaceuticals5', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals5', 'Major Pharmaceuticals4', 'Major Pharmaceuticals3', 'Major Pharmaceuticals3', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals5', 'Major Pharmaceuticals5', 'Major Pharmaceuticals5', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals5', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals3', 'Major Pharmaceuticals1', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals3', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals6', 'Major Pharmaceuticals4'], ['Newspapers/Magazines3', 'Newspapers/Magazines3', 'Newspapers/Magazines3', 'Newspapers/Magazines3', 'Newspapers/Magazines3', 'Newspapers/Magazines3', 'Newspapers/Magazines3', 'Newspapers/Magazines3', 'Newspapers/Magazines2', 'Newspapers/Magazines3']]
|
MIT
|
Stocks/Place Stock Trades into Senator Dataframe Ankur Edit.ipynb
|
paulmtree/Suspicious-Senator-Trading
|
Collaborative filtering on Google Analytics dataThis notebook demonstrates how to implement a WALS matrix refactorization approach to do collaborative filtering.
|
import os
PROJECT = "qwiklabs-gcp-00-34ffb0f0dc65" # REPLACE WITH YOUR PROJECT ID
BUCKET = "cloud-training-demos-ml" # REPLACE WITH YOUR BUCKET NAME
REGION = "us-central1" # REPLACE WITH YOUR BUCKET REGION e.g. us-central1
# Do not change these
os.environ["PROJECT"] = PROJECT
os.environ["BUCKET"] = BUCKET
os.environ["REGION"] = REGION
os.environ["TFVERSION"] = "1.13"
%%bash
gcloud config set project $PROJECT
gcloud config set compute/region $REGION
import tensorflow as tf
print(tf.__version__)
|
1.13.1
|
Apache-2.0
|
courses/machine_learning/deepdive/10_recommend/wals.ipynb
|
gozer/training-data-analyst
|
Create raw datasetFor collaborative filtering, we don't need to know anything about either the users or the content. Essentially, all we need to know is userId, itemId, and rating that the particular user gave the particular item.In this case, we are working with newspaper articles. The company doesn't ask their users to rate the articles. However, we can use the time-spent on the page as a proxy for rating.Normally, we would also add a time filter to this ("latest 7 days"), but our dataset is itself limited to a few days.
|
from google.cloud import bigquery
bq = bigquery.Client(project = PROJECT)
sql = """
#standardSQL
WITH CTE_visitor_page_content AS (
SELECT
fullVisitorID,
(SELECT MAX(IF(index=10, value, NULL)) FROM UNNEST(hits.customDimensions)) AS latestContentId,
(LEAD(hits.time, 1) OVER (PARTITION BY fullVisitorId ORDER BY hits.time ASC) - hits.time) AS session_duration
FROM
`cloud-training-demos.GA360_test.ga_sessions_sample`,
UNNEST(hits) AS hits
WHERE
# only include hits on pages
hits.type = "PAGE"
GROUP BY
fullVisitorId,
latestContentId,
hits.time )
-- Aggregate web stats
SELECT
fullVisitorID as visitorId,
latestContentId as contentId,
SUM(session_duration) AS session_duration
FROM
CTE_visitor_page_content
WHERE
latestContentId IS NOT NULL
GROUP BY
fullVisitorID,
latestContentId
HAVING
session_duration > 0
ORDER BY
latestContentId
"""
df = bq.query(sql).to_dataframe()
df.head()
stats = df.describe()
stats
df[["session_duration"]].plot(kind="hist", logy=True, bins=100, figsize=[8,5])
# The rating is the session_duration scaled to be in the range 0-1. This will help with training.
median = stats.loc["50%", "session_duration"]
df["rating"] = 0.3 * df["session_duration"] / median
df.loc[df["rating"] > 1, "rating"] = 1
df[["rating"]].plot(kind="hist", logy=True, bins=100, figsize=[8,5])
del df["session_duration"]
%%bash
rm -rf data
mkdir data
df.to_csv(path_or_buf = "data/collab_raw.csv", index = False, header = False)
!head data/collab_raw.csv
|
7337153711992174438,100074831,0.2321051400452234
5190801220865459604,100170790,1.0
2293633612703952721,100510126,0.2481776360816793
5874973374932455844,100510126,0.16690549004998828
1173698801255170595,100676857,0.05464232805149575
883397426232997550,10083328,0.9487035095774818
1808867070685560283,100906145,1.0
7615995624631762562,100906145,0.48418654214351925
5519169380728479914,100915139,0.20026163722525925
3427736932800080345,100950628,0.558924688331153
|
Apache-2.0
|
courses/machine_learning/deepdive/10_recommend/wals.ipynb
|
gozer/training-data-analyst
|
Create dataset for WALSThe raw dataset (above) won't work for WALS: The userId and itemId have to be 0,1,2 ... so we need to create a mapping from visitorId (in the raw data) to userId and contentId (in the raw data) to itemId. We will need to save the above mapping to a file because at prediction time, we'll need to know how to map the contentId in the table above to the itemId. We'll need two files: a "rows" dataset where all the items for a particular user are listed; and a "columns" dataset where all the users for a particular item are listed. Mapping
|
import pandas as pd
import numpy as np
def create_mapping(values, filename):
with open(filename, 'w') as ofp:
value_to_id = {value:idx for idx, value in enumerate(values.unique())}
for value, idx in value_to_id.items():
ofp.write("{},{}\n".format(value, idx))
return value_to_id
df = pd.read_csv(filepath_or_buffer = "data/collab_raw.csv",
header = None,
names = ["visitorId", "contentId", "rating"],
dtype = {"visitorId": str, "contentId": str, "rating": np.float})
df.to_csv(path_or_buf = "data/collab_raw.csv", index = False, header = False)
user_mapping = create_mapping(df["visitorId"], "data/users.csv")
item_mapping = create_mapping(df["contentId"], "data/items.csv")
!head -3 data/*.csv
df["userId"] = df["visitorId"].map(user_mapping.get)
df["itemId"] = df["contentId"].map(item_mapping.get)
mapped_df = df[["userId", "itemId", "rating"]]
mapped_df.to_csv(path_or_buf = "data/collab_mapped.csv", index = False, header = False)
mapped_df.head()
|
_____no_output_____
|
Apache-2.0
|
courses/machine_learning/deepdive/10_recommend/wals.ipynb
|
gozer/training-data-analyst
|
Creating rows and columns datasets
|
import pandas as pd
import numpy as np
mapped_df = pd.read_csv(filepath_or_buffer = "data/collab_mapped.csv", header = None, names = ["userId", "itemId", "rating"])
mapped_df.head()
NITEMS = np.max(mapped_df["itemId"]) + 1
NUSERS = np.max(mapped_df["userId"]) + 1
mapped_df["rating"] = np.round(mapped_df["rating"].values, 2)
print("{} items, {} users, {} interactions".format( NITEMS, NUSERS, len(mapped_df) ))
grouped_by_items = mapped_df.groupby("itemId")
iter = 0
for item, grouped in grouped_by_items:
print(item, grouped["userId"].values, grouped["rating"].values)
iter = iter + 1
if iter > 5:
break
import tensorflow as tf
grouped_by_items = mapped_df.groupby("itemId")
with tf.python_io.TFRecordWriter("data/users_for_item") as ofp:
for item, grouped in grouped_by_items:
example = tf.train.Example(features = tf.train.Features(feature = {
"key": tf.train.Feature(int64_list = tf.train.Int64List(value = [item])),
"indices": tf.train.Feature(int64_list = tf.train.Int64List(value = grouped["userId"].values)),
"values": tf.train.Feature(float_list = tf.train.FloatList(value = grouped["rating"].values))
}))
ofp.write(example.SerializeToString())
grouped_by_users = mapped_df.groupby("userId")
with tf.python_io.TFRecordWriter("data/items_for_user") as ofp:
for user, grouped in grouped_by_users:
example = tf.train.Example(features = tf.train.Features(feature = {
"key": tf.train.Feature(int64_list = tf.train.Int64List(value = [user])),
"indices": tf.train.Feature(int64_list = tf.train.Int64List(value = grouped["itemId"].values)),
"values": tf.train.Feature(float_list = tf.train.FloatList(value = grouped["rating"].values))
}))
ofp.write(example.SerializeToString())
!ls -lrt data
|
total 31908
-rw-r--r-- 1 jupyter jupyter 13152765 Jul 31 20:41 collab_raw.csv
-rw-r--r-- 1 jupyter jupyter 2134511 Jul 31 20:41 users.csv
-rw-r--r-- 1 jupyter jupyter 82947 Jul 31 20:41 items.csv
-rw-r--r-- 1 jupyter jupyter 7812739 Jul 31 20:41 collab_mapped.csv
-rw-r--r-- 1 jupyter jupyter 2252828 Jul 31 20:41 users_for_item
-rw-r--r-- 1 jupyter jupyter 7217822 Jul 31 20:41 items_for_user
|
Apache-2.0
|
courses/machine_learning/deepdive/10_recommend/wals.ipynb
|
gozer/training-data-analyst
|
To summarize, we created the following data files from collab_raw.csv: ```collab_mapped.csv``` is essentially the same data as in ```collab_raw.csv``` except that ```visitorId``` and ```contentId``` which are business-specific have been mapped to ```userId``` and ```itemId``` which are enumerated in 0,1,2,.... The mappings themselves are stored in ```items.csv``` and ```users.csv``` so that they can be used during inference. ```users_for_item``` contains all the users/ratings for each item in TFExample format ```items_for_user``` contains all the items/ratings for each user in TFExample format Train with WALSOnce you have the dataset, do matrix factorization with WALS using the [WALSMatrixFactorization](https://www.tensorflow.org/versions/master/api_docs/python/tf/contrib/factorization/WALSMatrixFactorization) in the contrib directory.This is an estimator model, so it should be relatively familiar.As usual, we write an input_fn to provide the data to the model, and then create the Estimator to do train_and_evaluate.Because it is in contrib and hasn't moved over to tf.estimator yet, we use tf.contrib.learn.Experiment to handle the training loop.
|
import os
import tensorflow as tf
from tensorflow.python.lib.io import file_io
from tensorflow.contrib.factorization import WALSMatrixFactorization
def read_dataset(mode, args):
def decode_example(protos, vocab_size):
features = {
"key": tf.FixedLenFeature(shape = [1], dtype = tf.int64),
"indices": tf.VarLenFeature(dtype = tf.int64),
"values": tf.VarLenFeature(dtype = tf.float32)}
parsed_features = tf.parse_single_example(serialized = protos, features = features)
values = tf.sparse_merge(sp_ids = parsed_features["indices"], sp_values = parsed_features["values"], vocab_size = vocab_size)
# Save key to remap after batching
# This is a temporary workaround to assign correct row numbers in each batch.
# You can ignore details of this part and remap_keys().
key = parsed_features["key"]
decoded_sparse_tensor = tf.SparseTensor(indices = tf.concat(values = [values.indices, [key]], axis = 0),
values = tf.concat(values = [values.values, [0.0]], axis = 0),
dense_shape = values.dense_shape)
return decoded_sparse_tensor
def remap_keys(sparse_tensor):
# Current indices of our SparseTensor that we need to fix
bad_indices = sparse_tensor.indices # shape = (current_batch_size * (number_of_items/users[i] + 1), 2)
# Current values of our SparseTensor that we need to fix
bad_values = sparse_tensor.values # shape = (current_batch_size * (number_of_items/users[i] + 1),)
# Since batch is ordered, the last value for a batch index is the user
# Find where the batch index chages to extract the user rows
# 1 where user, else 0
user_mask = tf.concat(values = [bad_indices[1:,0] - bad_indices[:-1,0], tf.constant(value = [1], dtype = tf.int64)], axis = 0) # shape = (current_batch_size * (number_of_items/users[i] + 1), 2)
# Mask out the user rows from the values
good_values = tf.boolean_mask(tensor = bad_values, mask = tf.equal(x = user_mask, y = 0)) # shape = (current_batch_size * number_of_items/users[i],)
item_indices = tf.boolean_mask(tensor = bad_indices, mask = tf.equal(x = user_mask, y = 0)) # shape = (current_batch_size * number_of_items/users[i],)
user_indices = tf.boolean_mask(tensor = bad_indices, mask = tf.equal(x = user_mask, y = 1))[:, 1] # shape = (current_batch_size,)
good_user_indices = tf.gather(params = user_indices, indices = item_indices[:,0]) # shape = (current_batch_size * number_of_items/users[i],)
# User and item indices are rank 1, need to make rank 1 to concat
good_user_indices_expanded = tf.expand_dims(input = good_user_indices, axis = -1) # shape = (current_batch_size * number_of_items/users[i], 1)
good_item_indices_expanded = tf.expand_dims(input = item_indices[:, 1], axis = -1) # shape = (current_batch_size * number_of_items/users[i], 1)
good_indices = tf.concat(values = [good_user_indices_expanded, good_item_indices_expanded], axis = 1) # shape = (current_batch_size * number_of_items/users[i], 2)
remapped_sparse_tensor = tf.SparseTensor(indices = good_indices, values = good_values, dense_shape = sparse_tensor.dense_shape)
return remapped_sparse_tensor
def parse_tfrecords(filename, vocab_size):
if mode == tf.estimator.ModeKeys.TRAIN:
num_epochs = None # indefinitely
else:
num_epochs = 1 # end-of-input after this
files = tf.gfile.Glob(filename = os.path.join(args["input_path"], filename))
# Create dataset from file list
dataset = tf.data.TFRecordDataset(files)
dataset = dataset.map(map_func = lambda x: decode_example(x, vocab_size))
dataset = dataset.repeat(count = num_epochs)
dataset = dataset.batch(batch_size = args["batch_size"])
dataset = dataset.map(map_func = lambda x: remap_keys(x))
return dataset.make_one_shot_iterator().get_next()
def _input_fn():
features = {
WALSMatrixFactorization.INPUT_ROWS: parse_tfrecords("items_for_user", args["nitems"]),
WALSMatrixFactorization.INPUT_COLS: parse_tfrecords("users_for_item", args["nusers"]),
WALSMatrixFactorization.PROJECT_ROW: tf.constant(True)
}
return features, None
return _input_fn
|
_____no_output_____
|
Apache-2.0
|
courses/machine_learning/deepdive/10_recommend/wals.ipynb
|
gozer/training-data-analyst
|
This code is helpful in developing the input function. You don't need it in production.
|
def try_out():
with tf.Session() as sess:
fn = read_dataset(
mode = tf.estimator.ModeKeys.EVAL,
args = {"input_path": "data", "batch_size": 4, "nitems": NITEMS, "nusers": NUSERS})
feats, _ = fn()
print(feats["input_rows"].eval())
print(feats["input_rows"].eval())
try_out()
def find_top_k(user, item_factors, k):
all_items = tf.matmul(a = tf.expand_dims(input = user, axis = 0), b = tf.transpose(a = item_factors))
topk = tf.nn.top_k(input = all_items, k = k)
return tf.cast(x = topk.indices, dtype = tf.int64)
def batch_predict(args):
import numpy as np
with tf.Session() as sess:
estimator = tf.contrib.factorization.WALSMatrixFactorization(
num_rows = args["nusers"],
num_cols = args["nitems"],
embedding_dimension = args["n_embeds"],
model_dir = args["output_dir"])
# This is how you would get the row factors for out-of-vocab user data
# row_factors = list(estimator.get_projections(input_fn=read_dataset(tf.estimator.ModeKeys.EVAL, args)))
# user_factors = tf.convert_to_tensor(np.array(row_factors))
# But for in-vocab data, the row factors are already in the checkpoint
user_factors = tf.convert_to_tensor(value = estimator.get_row_factors()[0]) # (nusers, nembeds)
# In either case, we have to assume catalog doesn"t change, so col_factors are read in
item_factors = tf.convert_to_tensor(value = estimator.get_col_factors()[0])# (nitems, nembeds)
# For each user, find the top K items
topk = tf.squeeze(input = tf.map_fn(fn = lambda user: find_top_k(user, item_factors, args["topk"]), elems = user_factors, dtype = tf.int64))
with file_io.FileIO(os.path.join(args["output_dir"], "batch_pred.txt"), mode = 'w') as f:
for best_items_for_user in topk.eval():
f.write(",".join(str(x) for x in best_items_for_user) + '\n')
def train_and_evaluate(args):
train_steps = int(0.5 + (1.0 * args["num_epochs"] * args["nusers"]) / args["batch_size"])
steps_in_epoch = int(0.5 + args["nusers"] / args["batch_size"])
print("Will train for {} steps, evaluating once every {} steps".format(train_steps, steps_in_epoch))
def experiment_fn(output_dir):
return tf.contrib.learn.Experiment(
tf.contrib.factorization.WALSMatrixFactorization(
num_rows = args["nusers"],
num_cols = args["nitems"],
embedding_dimension = args["n_embeds"],
model_dir = args["output_dir"]),
train_input_fn = read_dataset(tf.estimator.ModeKeys.TRAIN, args),
eval_input_fn = read_dataset(tf.estimator.ModeKeys.EVAL, args),
train_steps = train_steps,
eval_steps = 1,
min_eval_frequency = steps_in_epoch
)
from tensorflow.contrib.learn.python.learn import learn_runner
learn_runner.run(experiment_fn = experiment_fn, output_dir = args["output_dir"])
batch_predict(args)
import shutil
shutil.rmtree(path = "wals_trained", ignore_errors=True)
train_and_evaluate({
"output_dir": "wals_trained",
"input_path": "data/",
"num_epochs": 0.05,
"nitems": NITEMS,
"nusers": NUSERS,
"batch_size": 512,
"n_embeds": 10,
"topk": 3
})
!ls wals_trained
!head wals_trained/batch_pred.txt
|
284,5609,36
284,2754,42
284,3168,534
2621,5528,2694
4409,5295,343
5161,3267,3369
5479,1335,55
5479,1335,55
4414,284,5572
284,241,2359
|
Apache-2.0
|
courses/machine_learning/deepdive/10_recommend/wals.ipynb
|
gozer/training-data-analyst
|
Run as a Python moduleLet's run it as Python module for just a few steps.
|
os.environ["NITEMS"] = str(NITEMS)
os.environ["NUSERS"] = str(NUSERS)
%%bash
rm -rf wals.tar.gz wals_trained
gcloud ml-engine local train \
--module-name=walsmodel.task \
--package-path=${PWD}/walsmodel \
-- \
--output_dir=${PWD}/wals_trained \
--input_path=${PWD}/data \
--num_epochs=0.01 --nitems=${NITEMS} --nusers=${NUSERS} \
--job-dir=./tmp
|
Will train for 2 steps, evaluating once every 162 steps
|
Apache-2.0
|
courses/machine_learning/deepdive/10_recommend/wals.ipynb
|
gozer/training-data-analyst
|
Run on Cloud
|
%%bash
gsutil -m cp data/* gs://${BUCKET}/wals/data
%%bash
OUTDIR=gs://${BUCKET}/wals/model_trained
JOBNAME=wals_$(date -u +%y%m%d_%H%M%S)
echo $OUTDIR $REGION $JOBNAME
gsutil -m rm -rf $OUTDIR
gcloud ml-engine jobs submit training $JOBNAME \
--region=$REGION \
--module-name=walsmodel.task \
--package-path=${PWD}/walsmodel \
--job-dir=$OUTDIR \
--staging-bucket=gs://$BUCKET \
--scale-tier=BASIC_GPU \
--runtime-version=$TFVERSION \
-- \
--output_dir=$OUTDIR \
--input_path=gs://${BUCKET}/wals/data \
--num_epochs=10 --nitems=${NITEMS} --nusers=${NUSERS}
|
_____no_output_____
|
Apache-2.0
|
courses/machine_learning/deepdive/10_recommend/wals.ipynb
|
gozer/training-data-analyst
|
This took 10 minutes for me. Get row and column factorsOnce you have a trained WALS model, you can get row and column factors (user and item embeddings) from the checkpoint file. We'll look at how to use these in the section on building a recommendation system using deep neural networks.
|
def get_factors(args):
with tf.Session() as sess:
estimator = tf.contrib.factorization.WALSMatrixFactorization(
num_rows = args["nusers"],
num_cols = args["nitems"],
embedding_dimension = args["n_embeds"],
model_dir = args["output_dir"])
row_factors = estimator.get_row_factors()[0]
col_factors = estimator.get_col_factors()[0]
return row_factors, col_factors
args = {
"output_dir": "gs://{}/wals/model_trained".format(BUCKET),
"nitems": NITEMS,
"nusers": NUSERS,
"n_embeds": 10
}
user_embeddings, item_embeddings = get_factors(args)
print(user_embeddings[:3])
print(item_embeddings[:3])
|
INFO:tensorflow:Using default config.
INFO:tensorflow:Using config: {'_environment': 'local', '_is_chief': True, '_keep_checkpoint_every_n_hours': 10000, '_num_worker_replicas': 0, '_session_config': None, '_task_type': None, '_eval_distribute': None, '_tf_config': gpu_options {
per_process_gpu_memory_fraction: 1.0
}
, '_master': '', '_log_step_count_steps': 100, '_model_dir': 'gs://qwiklabs-gcp-cbc8684b07fc2dbd-bucket/wals/model_trained', '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x7f4bd8302f28>, '_device_fn': None, '_keep_checkpoint_max': 5, '_task_id': 0, '_evaluation_master': '', '_save_checkpoints_steps': None, '_protocol': None, '_train_distribute': None, '_save_checkpoints_secs': 600, '_save_summary_steps': 100, '_tf_random_seed': None, '_num_ps_replicas': 0}
[[ 3.3451824e-06 -1.1986867e-05 4.8447573e-06 -1.5209486e-05
-1.7004859e-07 1.1976428e-05 9.8887876e-06 7.2386983e-06
-7.0237149e-07 -7.9796819e-06]
[-2.5300323e-03 1.4055537e-03 -9.8291773e-04 -4.2533795e-03
-1.4166030e-03 -1.9530674e-03 8.5932651e-04 -1.5276540e-03
2.1342330e-03 1.2041229e-03]
[ 9.5228699e-21 5.5453966e-21 2.2947056e-21 -5.8859543e-21
7.7516509e-21 -2.7640896e-20 2.3587296e-20 -3.9876822e-21
1.7312470e-20 2.5409211e-20]]
[[-1.2125404e-06 -8.6304914e-05 4.4657736e-05 -6.8423047e-05
5.8551927e-06 9.7241784e-05 6.6776753e-05 1.6673854e-05
-1.2708440e-05 -5.1148414e-05]
[-1.1353870e-01 5.9097271e-02 -4.6105500e-02 -1.5460028e-01
-1.9166643e-02 -7.3236257e-02 3.5582058e-02 -5.6805085e-02
7.5831160e-02 7.5306065e-02]
[ 7.1989548e-20 4.4574543e-20 6.5149121e-21 -4.6291777e-20
8.8196718e-20 -2.3245078e-19 1.9459292e-19 4.0191465e-20
1.6273659e-19 2.2836562e-19]]
|
Apache-2.0
|
courses/machine_learning/deepdive/10_recommend/wals.ipynb
|
gozer/training-data-analyst
|
You can visualize the embedding vectors using dimensional reduction techniques such as PCA.
|
import matplotlib.pyplot as plt
from mpl_toolkits.mplot3d import Axes3D
from sklearn.decomposition import PCA
pca = PCA(n_components = 3)
pca.fit(user_embeddings)
user_embeddings_pca = pca.transform(user_embeddings)
fig = plt.figure(figsize = (8,8))
ax = fig.add_subplot(111, projection = "3d")
xs, ys, zs = user_embeddings_pca[::150].T
ax.scatter(xs, ys, zs)
|
_____no_output_____
|
Apache-2.0
|
courses/machine_learning/deepdive/10_recommend/wals.ipynb
|
gozer/training-data-analyst
|
window.dataLayer = window.dataLayer || []; function gtag(){dataLayer.push(arguments);} gtag('js', new Date()); gtag('config', 'UA-59152712-8'); Computing the 4-Velocity Time-Component $u^0$, the Magnetic Field Measured by a Comoving Observer $b^{\mu}$, and the Poynting Vector $S^i$ Authors: Zach Etienne & Patrick Nelson[comment]: (Abstract: TODO)**Notebook Status:** Validated **Validation Notes:** This module has been validated against a trusted code (the hand-written smallbPoynET in WVUThorns_diagnostics, which itself is based on expressions in IllinoisGRMHD... which was validated against the original GRMHD code of the Illinois NR group) NRPy+ Source Code for this module: [u0_smallb_Poynting__Cartesian.py](../edit/u0_smallb_Poynting__Cartesian/u0_smallb_Poynting__Cartesian.py)[comment]: (Introduction: TODO) Table of Contents$$\label{toc}$$This notebook is organized as follows1. [Step 1](u0bu): Computing $u^0$ and $b^{\mu}$ 1. [Step 1.a](4metric): Compute the 4-metric $g_{\mu\nu}$ and its inverse $g^{\mu\nu}$ from the ADM 3+1 variables, using the [`BSSN.ADMBSSN_tofrom_4metric`](../edit/BSSN/ADMBSSN_tofrom_4metric.py) ([**tutorial**](Tutorial-ADMBSSN_tofrom_4metric.ipynb)) NRPy+ module 1. [Step 1.b](u0): Compute $u^0$ from the Valencia 3-velocity 1. [Step 1.c](uj): Compute $u_j$ from $u^0$, the Valencia 3-velocity, and $g_{\mu\nu}$ 1. [Step 1.d](gamma): Compute $\gamma=$ `gammaDET` from the ADM 3+1 variables 1. [Step 1.e](beta): Compute $b^\mu$1. [Step 2](poynting_flux): Defining the Poynting Flux Vector $S^{i}$ 1. [Step 2.a](g): Computing $g^{i\nu}$ 1. [Step 2.b](s): Computing $S^{i}$1. [Step 3](code_validation): Code Validation against `u0_smallb_Poynting__Cartesian` NRPy+ module1. [Step 4](appendix): Appendix: Proving Eqs. 53 and 56 in [Duez *et al* (2005)](https://arxiv.org/pdf/astro-ph/0503420.pdf)1. [Step 5](latex_pdf_output): Output this notebook to $\LaTeX$-formatted PDF file Step 1: Computing $u^0$ and $b^{\mu}$ \[Back to [top](toc)\]$$\label{u0bu}$$First some definitions. The spatial components of $b^{\mu}$ are simply the magnetic field as measured by an observer comoving with the plasma $B^{\mu}_{\rm (u)}$, divided by $\sqrt{4\pi}$. In addition, in the ideal MHD limit, $B^{\mu}_{\rm (u)}$ is orthogonal to the plasma 4-velocity $u^\mu$, which sets the $\mu=0$ component. Note also that $B^{\mu}_{\rm (u)}$ is related to the magnetic field as measured by a *normal* observer $B^i$ via a simple projection (Eq 21 in [Duez *et al* (2005)](https://arxiv.org/pdf/astro-ph/0503420.pdf)), which results in the expressions (Eqs 23 and 24 in [Duez *et al* (2005)](https://arxiv.org/pdf/astro-ph/0503420.pdf)):\begin{align}\sqrt{4\pi} b^0 = B^0_{\rm (u)} &= \frac{u_j B^j}{\alpha} \\\sqrt{4\pi} b^i = B^i_{\rm (u)} &= \frac{B^i + (u_j B^j) u^i}{\alpha u^0}\\\end{align}$B^i$ is related to the actual magnetic field evaluated in IllinoisGRMHD, $\tilde{B}^i$ via$$B^i = \frac{\tilde{B}^i}{\gamma},$$where $\gamma$ is the determinant of the spatial 3-metric.The above expressions will require that we compute1. the 4-metric $g_{\mu\nu}$ from the ADM 3+1 variables1. $u^0$ from the Valencia 3-velocity1. $u_j$ from $u^0$, the Valencia 3-velocity, and $g_{\mu\nu}$1. $\gamma$ from the ADM 3+1 variables Step 1.a: Compute the 4-metric $g_{\mu\nu}$ and its inverse $g^{\mu\nu}$ from the ADM 3+1 variables, using the [`BSSN.ADMBSSN_tofrom_4metric`](../edit/BSSN/ADMBSSN_tofrom_4metric.py) ([**tutorial**](Tutorial-ADMBSSN_tofrom_4metric.ipynb)) NRPy+ module \[Back to [top](toc)\]$$\label{4metric}$$We are given $\gamma_{ij}$, $\alpha$, and $\beta^i$ from ADMBase, so let's first compute $$g_{\mu\nu} = \begin{pmatrix} -\alpha^2 + \beta^k \beta_k & \beta_i \\\beta_j & \gamma_{ij}\end{pmatrix}.$$
|
# Step 1: Initialize needed Python/NRPy+ modules
import sympy as sp # SymPy: The Python computer algebra package upon which NRPy+ depends
import NRPy_param_funcs as par # NRPy+: Parameter interface
import indexedexp as ixp # NRPy+: Symbolic indexed expression (e.g., tensors, vectors, etc.) support
import reference_metric as rfm # NRPy+: Reference metric support
from outputC import * # NRPy+: Basic C code output functionality
import BSSN.ADMBSSN_tofrom_4metric as AB4m # NRPy+: ADM/BSSN <-> 4-metric conversions
# Set spatial dimension = 3
DIM=3
thismodule = "smallbPoynET"
# Step 1.a: Compute the 4-metric $g_{\mu\nu}$ and its inverse
# $g^{\mu\nu}$ from the ADM 3+1 variables, using the
# BSSN.ADMBSSN_tofrom_4metric NRPy+ module
import BSSN.ADMBSSN_tofrom_4metric as AB4m
gammaDD,betaU,alpha = AB4m.setup_ADM_quantities("ADM")
AB4m.g4DD_ito_BSSN_or_ADM("ADM",gammaDD,betaU,alpha)
g4DD = AB4m.g4DD
AB4m.g4UU_ito_BSSN_or_ADM("ADM",gammaDD,betaU,alpha)
g4UU = AB4m.g4UU
|
_____no_output_____
|
BSD-2-Clause
|
Tutorial-u0_smallb_Poynting-Cartesian.ipynb
|
KAClough/nrpytutorial
|
Step 1.b: Compute $u^0$ from the Valencia 3-velocity \[Back to [top](toc)\]$$\label{u0}$$According to Eqs. 9-11 of [the IllinoisGRMHD paper](https://arxiv.org/pdf/1501.07276.pdf), the Valencia 3-velocity $v^i_{(n)}$ is related to the 4-velocity $u^\mu$ via\begin{align}\alpha v^i_{(n)} &= \frac{u^i}{u^0} + \beta^i \\\implies u^i &= u^0 \left(\alpha v^i_{(n)} - \beta^i\right)\end{align}Defining $v^i = \frac{u^i}{u^0}$, we get$$v^i = \alpha v^i_{(n)} - \beta^i,$$and in terms of this variable we get\begin{align}g_{00} \left(u^0\right)^2 + 2 g_{0i} u^0 u^i + g_{ij} u^i u^j &= \left(u^0\right)^2 \left(g_{00} + 2 g_{0i} v^i + g_{ij} v^i v^j\right)\\\implies u^0 &= \pm \sqrt{\frac{-1}{g_{00} + 2 g_{0i} v^i + g_{ij} v^i v^j}} \\&= \pm \sqrt{\frac{-1}{(-\alpha^2 + \beta^2) + 2 \beta_i v^i + \gamma_{ij} v^i v^j}} \\&= \pm \sqrt{\frac{1}{\alpha^2 - \gamma_{ij}\left(\beta^i + v^i\right)\left(\beta^j + v^j\right)}}\\&= \pm \sqrt{\frac{1}{\alpha^2 - \alpha^2 \gamma_{ij}v^i_{(n)}v^j_{(n)}}}\\&= \pm \frac{1}{\alpha}\sqrt{\frac{1}{1 - \gamma_{ij}v^i_{(n)}v^j_{(n)}}}\end{align}Generally speaking, numerical errors will occasionally drive expressions under the radical to either negative values or potentially enormous values (corresponding to enormous Lorentz factors). Thus a reliable approach for computing $u^0$ requires that we first rewrite the above expression in terms of the Lorentz factor squared: $\Gamma^2=\left(\alpha u^0\right)^2$:\begin{align}u^0 &= \pm \frac{1}{\alpha}\sqrt{\frac{1}{1 - \gamma_{ij}v^i_{(n)}v^j_{(n)}}}\\\implies \left(\alpha u^0\right)^2 &= \frac{1}{1 - \gamma_{ij}v^i_{(n)}v^j_{(n)}} \\\implies \gamma_{ij}v^i_{(n)}v^j_{(n)} &= 1 - \frac{1}{\left(\alpha u^0\right)^2} \\&= 1 - \frac{1}{\Gamma^2}\end{align}In order for the bottom expression to hold true, the left-hand side must be between 0 and 1. Again, this is not guaranteed due to the appearance of numerical errors. In fact, a robust algorithm will not allow $\Gamma^2$ to become too large (which might contribute greatly to the stress-energy of a given gridpoint), so let's define $\Gamma_{\rm max}$, the largest allowed Lorentz factor. Then our algorithm for computing $u^0$ is as follows:If$$R=\gamma_{ij}v^i_{(n)}v^j_{(n)}>1 - \frac{1}{\Gamma_{\rm max}^2},$$ then adjust the 3-velocity $v^i$ as follows:$$v^i_{(n)} = \sqrt{\frac{1 - \frac{1}{\Gamma_{\rm max}^2}}{R}}v^i_{(n)}.$$After this rescaling, we are then guaranteed that if $R$ is recomputed, it will be set to its ceiling value $R=R_{\rm max} = 1 - \frac{1}{\Gamma_{\rm max}^2}$.Then, regardless of whether the ceiling on $R$ was applied, $u^0$ can be safely computed via$$u^0 = \frac{1}{\alpha \sqrt{1-R}}.$$
|
ValenciavU = ixp.register_gridfunctions_for_single_rank1("AUX","ValenciavU",DIM=3)
# Step 1: Compute R = 1 - 1/max(Gamma)
R = sp.sympify(0)
for i in range(DIM):
for j in range(DIM):
R += gammaDD[i][j]*ValenciavU[i]*ValenciavU[j]
GAMMA_SPEED_LIMIT = par.Cparameters("REAL",thismodule,"GAMMA_SPEED_LIMIT",10.0) # Default value based on
# IllinoisGRMHD.
# GiRaFFE default = 2000.0
Rmax = 1 - 1/(GAMMA_SPEED_LIMIT*GAMMA_SPEED_LIMIT)
rescaledValenciavU = ixp.zerorank1()
for i in range(DIM):
rescaledValenciavU[i] = ValenciavU[i]*sp.sqrt(Rmax/R)
rescaledu0 = 1/(alpha*sp.sqrt(1-Rmax))
regularu0 = 1/(alpha*sp.sqrt(1-R))
computeu0_Cfunction = """
/* Function for computing u^0 from Valencia 3-velocity. */
/* Inputs: ValenciavU[], alpha, gammaDD[][], GAMMA_SPEED_LIMIT (C parameter) */
/* Output: u0=u^0 and velocity-limited ValenciavU[] */\n\n"""
computeu0_Cfunction += outputC([R,Rmax],["const double R","const double Rmax"],"returnstring",
params="includebraces=False,CSE_varprefix=tmpR,outCverbose=False")
computeu0_Cfunction += "if(R <= Rmax) "
computeu0_Cfunction += outputC(regularu0,"u0","returnstring",
params="includebraces=True,CSE_varprefix=tmpnorescale,outCverbose=False")
computeu0_Cfunction += " else "
computeu0_Cfunction += outputC([rescaledValenciavU[0],rescaledValenciavU[1],rescaledValenciavU[2],rescaledu0],
["ValenciavU0","ValenciavU1","ValenciavU2","u0"],"returnstring",
params="includebraces=True,CSE_varprefix=tmprescale,outCverbose=False")
print(computeu0_Cfunction)
|
/* Function for computing u^0 from Valencia 3-velocity. */
/* Inputs: ValenciavU[], alpha, gammaDD[][], GAMMA_SPEED_LIMIT (C parameter) */
/* Output: u0=u^0 and velocity-limited ValenciavU[] */
const double tmpR0 = 2*ValenciavU0;
const double R = ((ValenciavU0)*(ValenciavU0))*gammaDD00 + ((ValenciavU1)*(ValenciavU1))*gammaDD11 + 2*ValenciavU1*ValenciavU2*gammaDD12 + ValenciavU1*gammaDD01*tmpR0 + ((ValenciavU2)*(ValenciavU2))*gammaDD22 + ValenciavU2*gammaDD02*tmpR0;
const double Rmax = 1 - 1/((GAMMA_SPEED_LIMIT)*(GAMMA_SPEED_LIMIT));
if(R <= Rmax) {
const double tmpnorescale0 = 2*ValenciavU0;
u0 = 1/(alpha*sqrt(-((ValenciavU0)*(ValenciavU0))*gammaDD00 - ((ValenciavU1)*(ValenciavU1))*gammaDD11 - 2*ValenciavU1*ValenciavU2*gammaDD12 - ValenciavU1*gammaDD01*tmpnorescale0 - ((ValenciavU2)*(ValenciavU2))*gammaDD22 - ValenciavU2*gammaDD02*tmpnorescale0 + 1));
}
else {
const double tmprescale0 = 2*ValenciavU0;
const double tmprescale1 = sqrt((1 - 1/((GAMMA_SPEED_LIMIT)*(GAMMA_SPEED_LIMIT)))/(((ValenciavU0)*(ValenciavU0))*gammaDD00 + ((ValenciavU1)*(ValenciavU1))*gammaDD11 + 2*ValenciavU1*ValenciavU2*gammaDD12 + ValenciavU1*gammaDD01*tmprescale0 + ((ValenciavU2)*(ValenciavU2))*gammaDD22 + ValenciavU2*gammaDD02*tmprescale0));
ValenciavU0 = ValenciavU0*tmprescale1;
ValenciavU1 = ValenciavU1*tmprescale1;
ValenciavU2 = ValenciavU2*tmprescale1;
u0 = fabs(GAMMA_SPEED_LIMIT)/alpha;
}
|
BSD-2-Clause
|
Tutorial-u0_smallb_Poynting-Cartesian.ipynb
|
KAClough/nrpytutorial
|
Step 1.c: Compute $u_j$ from $u^0$, the Valencia 3-velocity, and $g_{\mu\nu}$ \[Back to [top](toc)\]$$\label{uj}$$The basic equation is\begin{align}u_j &= g_{\mu j} u^{\mu} \\&= g_{0j} u^0 + g_{ij} u^i \\&= \beta_j u^0 + \gamma_{ij} u^i \\&= \beta_j u^0 + \gamma_{ij} u^0 \left(\alpha v^i_{(n)} - \beta^i\right) \\&= u^0 \left(\beta_j + \gamma_{ij} \left(\alpha v^i_{(n)} - \beta^i\right) \right)\\&= \alpha u^0 \gamma_{ij} v^i_{(n)} \\\end{align}
|
u0 = par.Cparameters("REAL",thismodule,"u0",1e300) # Will be overwritten in C code. Set to crazy value to ensure this.
uD = ixp.zerorank1()
for i in range(DIM):
for j in range(DIM):
uD[j] += alpha*u0*gammaDD[i][j]*ValenciavU[i]
|
_____no_output_____
|
BSD-2-Clause
|
Tutorial-u0_smallb_Poynting-Cartesian.ipynb
|
KAClough/nrpytutorial
|
Step 1.d: Compute $b^\mu$ \[Back to [top](toc)\]$$\label{beta}$$We compute $b^\mu$ from the above expressions:\begin{align}\sqrt{4\pi} b^0 = B^0_{\rm (u)} &= \frac{u_j B^j}{\alpha} \\\sqrt{4\pi} b^i = B^i_{\rm (u)} &= \frac{B^i + (u_j B^j) u^i}{\alpha u^0}\\\end{align}$B^i$ is exactly equal to the $B^i$ evaluated in IllinoisGRMHD/GiRaFFE.Pulling this together, we currently have available as input:+ $\tilde{B}^i$+ $u_j$+ $u^0$,with the goal of outputting now $b^\mu$ and $b^2$:
|
M_PI = par.Cparameters("#define",thismodule,"M_PI","")
BU = ixp.register_gridfunctions_for_single_rank1("AUX","BU",DIM=3)
# uBcontraction = u_i B^i
uBcontraction = sp.sympify(0)
for i in range(DIM):
uBcontraction += uD[i]*BU[i]
# uU = 3-vector representing u^i = u^0 \left(\alpha v^i_{(n)} - \beta^i\right)
uU = ixp.zerorank1()
for i in range(DIM):
uU[i] = u0*(alpha*ValenciavU[i] - betaU[i])
smallb4U = ixp.zerorank1(DIM=4)
smallb4U[0] = uBcontraction/(alpha*sp.sqrt(4*M_PI))
for i in range(DIM):
smallb4U[1+i] = (BU[i] + uBcontraction*uU[i])/(alpha*u0*sp.sqrt(4*M_PI))
|
_____no_output_____
|
BSD-2-Clause
|
Tutorial-u0_smallb_Poynting-Cartesian.ipynb
|
KAClough/nrpytutorial
|
Step 2: Defining the Poynting Flux Vector $S^{i}$ \[Back to [top](toc)\]$$\label{poynting_flux}$$The Poynting flux is defined in Eq. 11 of [Kelly *et al.*](https://arxiv.org/pdf/1710.02132.pdf) (note that we choose the minus sign convention so that the Poynting luminosity across a spherical shell is $L_{\rm EM} = \int (-\alpha T^i_{\rm EM\ 0}) \sqrt{\gamma} d\Omega = \int S^r \sqrt{\gamma} d\Omega$, as in [Farris *et al.*](https://arxiv.org/pdf/1207.3354.pdf):$$S^i = -\alpha T^i_{\rm EM\ 0} = -\alpha\left(b^2 u^i u_0 + \frac{1}{2} b^2 g^i{}_0 - b^i b_0\right)$$ Step 2.a: Computing $S^{i}$ \[Back to [top](toc)\]$$\label{s}$$Given $g^{\mu\nu}$ computed above, we focus first on the $g^i{}_{0}$ term by computing $$g^\mu{}_\delta = g^{\mu\nu} g_{\nu \delta},$$and then the rest of the Poynting flux vector can be immediately computed from quantities defined above:$$S^i = -\alpha T^i_{\rm EM\ 0} = -\alpha\left(b^2 u^i u_0 + \frac{1}{2} b^2 g^i{}_0 - b^i b_0\right)$$
|
# Step 2.a.i: compute g^\mu_\delta:
g4UD = ixp.zerorank2(DIM=4)
for mu in range(4):
for delta in range(4):
for nu in range(4):
g4UD[mu][delta] += g4UU[mu][nu]*g4DD[nu][delta]
# Step 2.a.ii: compute b_{\mu}
smallb4D = ixp.zerorank1(DIM=4)
for mu in range(4):
for nu in range(4):
smallb4D[mu] += g4DD[mu][nu]*smallb4U[nu]
# Step 2.a.iii: compute u_0 = g_{mu 0} u^{mu} = g4DD[0][0]*u0 + g4DD[i][0]*uU[i]
u_0 = g4DD[0][0]*u0
for i in range(DIM):
u_0 += g4DD[i+1][0]*uU[i]
# Step 2.a.iv: compute b^2, setting b^2 = smallb2etk, as gridfunctions with base names ending in a digit
# are forbidden in NRPy+.
smallb2etk = sp.sympify(0)
for mu in range(4):
smallb2etk += smallb4U[mu]*smallb4D[mu]
# Step 2.a.v: compute S^i
PoynSU = ixp.zerorank1()
for i in range(DIM):
PoynSU[i] = -alpha * (smallb2etk*uU[i]*u_0 + sp.Rational(1,2)*smallb2etk*g4UD[i+1][0] - smallb4U[i+1]*smallb4D[0])
|
_____no_output_____
|
BSD-2-Clause
|
Tutorial-u0_smallb_Poynting-Cartesian.ipynb
|
KAClough/nrpytutorial
|
Step 3: Code Validation against `u0_smallb_Poynting__Cartesian` NRPy+ module \[Back to [top](toc)\]$$\label{code_validation}$$Here, as a code validation check, we verify agreement in the SymPy expressions for u0, smallbU, smallb2etk, and PoynSU between1. this tutorial and 2. the NRPy+ [u0_smallb_Poynting__Cartesian module](../edit/u0_smallb_Poynting__Cartesian/u0_smallb_Poynting__Cartesian.py).
|
import sys
import u0_smallb_Poynting__Cartesian.u0_smallb_Poynting__Cartesian as u0etc
u0etc.compute_u0_smallb_Poynting__Cartesian(gammaDD,betaU,alpha,ValenciavU,BU)
if u0etc.computeu0_Cfunction != computeu0_Cfunction:
print("FAILURE: u0 C code has changed!")
sys.exit(1)
else:
print("PASSED: u0 C code matches!")
for i in range(4):
print("u0etc.smallb4U["+str(i)+"] - smallb4U["+str(i)+"] = "
+ str(u0etc.smallb4U[i]-smallb4U[i]))
print("u0etc.smallb2etk - smallb2etk = " + str(u0etc.smallb2etk-smallb2etk))
for i in range(DIM):
print("u0etc.PoynSU["+str(i)+"] - PoynSU["+str(i)+"] = "
+ str(u0etc.PoynSU[i]-PoynSU[i]))
|
PASSED: u0 C code matches!
u0etc.smallb4U[0] - smallb4U[0] = 0
u0etc.smallb4U[1] - smallb4U[1] = 0
u0etc.smallb4U[2] - smallb4U[2] = 0
u0etc.smallb4U[3] - smallb4U[3] = 0
u0etc.smallb2etk - smallb2etk = 0
u0etc.PoynSU[0] - PoynSU[0] = 0
u0etc.PoynSU[1] - PoynSU[1] = 0
u0etc.PoynSU[2] - PoynSU[2] = 0
|
BSD-2-Clause
|
Tutorial-u0_smallb_Poynting-Cartesian.ipynb
|
KAClough/nrpytutorial
|
Step 4: Appendix: Proving Eqs. 53 and 56 in [Duez *et al* (2005)](https://arxiv.org/pdf/astro-ph/0503420.pdf)$$\label{appendix}$$$u^\mu u_\mu = -1$ implies\begin{align}g^{\mu\nu} u_\mu u_\nu &= g^{00} \left(u_0\right)^2 + 2 g^{0i} u_0 u_i + g^{ij} u_i u_j = -1 \\\implies &g^{00} \left(u_0\right)^2 + 2 g^{0i} u_0 u_i + g^{ij} u_i u_j + 1 = 0\\& a x^2 + b x + c = 0\end{align}Thus we have a quadratic equation for $u_0$, with solution given by\begin{align}u_0 &= \frac{-b \pm \sqrt{b^2 - 4 a c}}{2 a} \\&= \frac{-2 g^{0i}u_i \pm \sqrt{\left(2 g^{0i} u_i\right)^2 - 4 g^{00} (g^{ij} u_i u_j + 1)}}{2 g^{00}}\\&= \frac{-g^{0i}u_i \pm \sqrt{\left(g^{0i} u_i\right)^2 - g^{00} (g^{ij} u_i u_j + 1)}}{g^{00}}\\\end{align}Notice that (Eq. 4.49 in [Gourgoulhon](https://arxiv.org/pdf/gr-qc/0703035.pdf))$$g^{\mu\nu} = \begin{pmatrix} -\frac{1}{\alpha^2} & \frac{\beta^i}{\alpha^2} \\\frac{\beta^i}{\alpha^2} & \gamma^{ij} - \frac{\beta^i\beta^j}{\alpha^2}\end{pmatrix},$$so we have\begin{align}u_0 &= \frac{-\beta^i u_i/\alpha^2 \pm \sqrt{\left(\beta^i u_i/\alpha^2\right)^2 + 1/\alpha^2 (g^{ij} u_i u_j + 1)}}{1/\alpha^2}\\&= -\beta^i u_i \pm \sqrt{\left(\beta^i u_i\right)^2 + \alpha^2 (g^{ij} u_i u_j + 1)}\\&= -\beta^i u_i \pm \sqrt{\left(\beta^i u_i\right)^2 + \alpha^2 \left(\left[\gamma^{ij} - \frac{\beta^i\beta^j}{\alpha^2}\right] u_i u_j + 1\right)}\\&= -\beta^i u_i \pm \sqrt{\left(\beta^i u_i\right)^2 + \alpha^2 \left(\gamma^{ij}u_i u_j + 1\right) - \beta^i\beta^j u_i u_j}\\&= -\beta^i u_i \pm \sqrt{\alpha^2 \left(\gamma^{ij}u_i u_j + 1\right)}\\\end{align}Now, since $$u^0 = g^{\alpha 0} u_\alpha = -\frac{1}{\alpha^2} u_0 + \frac{\beta^i u_i}{\alpha^2},$$we get\begin{align}u^0 &= \frac{1}{\alpha^2} \left(u_0 + \beta^i u_i\right) \\&= \pm \frac{1}{\alpha^2} \sqrt{\alpha^2 \left(\gamma^{ij}u_i u_j + 1\right)}\\&= \pm \frac{1}{\alpha} \sqrt{\gamma^{ij}u_i u_j + 1}\\\end{align}By convention, the relativistic Gamma factor is positive and given by $\alpha u^0$, so we choose the positive root. Thus we have derived Eq. 53 in [Duez *et al* (2005)](https://arxiv.org/pdf/astro-ph/0503420.pdf):$$u^0 = \frac{1}{\alpha} \sqrt{\gamma^{ij}u_i u_j + 1}.$$Next we evaluate \begin{align}u^i &= u_\mu g^{\mu i} \\&= u_0 g^{0 i} + u_j g^{i j}\\&= u_0 \frac{\beta^i}{\alpha^2} + u_j \left(\gamma^{ij} - \frac{\beta^i\beta^j}{\alpha^2}\right)\\&= \gamma^{ij} u_j + u_0 \frac{\beta^i}{\alpha^2} - u_j \frac{\beta^i\beta^j}{\alpha^2}\\&= \gamma^{ij} u_j + \frac{\beta^i}{\alpha^2} \left(u_0 - u_j \beta^j\right)\\&= \gamma^{ij} u_j - \beta^i u^0,\\\implies v^i &= \frac{\gamma^{ij} u_j}{u^0} - \beta^i\end{align}which is equivalent to Eq. 56 in [Duez *et al* (2005)](https://arxiv.org/pdf/astro-ph/0503420.pdf). Notice in the last step, we used the above definition of $u^0$. Step 5: Output this notebook to $\LaTeX$-formatted PDF file \[Back to [top](toc)\]$$\label{latex_pdf_output}$$The following code cell converts this Jupyter notebook into a proper, clickable $\LaTeX$-formatted PDF file. After the cell is successfully run, the generated PDF may be found in the root NRPy+ tutorial directory, with filename[Tutorial-u0_smallb_Poynting-Cartesian.pdf](Tutorial-u0_smallb_Poynting-Cartesian.pdf) (Note that clicking on this link may not work; you may need to open the PDF file through another means.)
|
!jupyter nbconvert --to latex --template latex_nrpy_style.tplx --log-level='WARN' Tutorial-u0_smallb_Poynting-Cartesian.ipynb
!pdflatex -interaction=batchmode Tutorial-u0_smallb_Poynting-Cartesian.tex
!pdflatex -interaction=batchmode Tutorial-u0_smallb_Poynting-Cartesian.tex
!pdflatex -interaction=batchmode Tutorial-u0_smallb_Poynting-Cartesian.tex
!rm -f Tut*.out Tut*.aux Tut*.log
|
[pandoc warning] Duplicate link reference `[comment]' "source" (line 22, column 1)
This is pdfTeX, Version 3.14159265-2.6-1.40.18 (TeX Live 2017/Debian) (preloaded format=pdflatex)
restricted \write18 enabled.
entering extended mode
This is pdfTeX, Version 3.14159265-2.6-1.40.18 (TeX Live 2017/Debian) (preloaded format=pdflatex)
restricted \write18 enabled.
entering extended mode
This is pdfTeX, Version 3.14159265-2.6-1.40.18 (TeX Live 2017/Debian) (preloaded format=pdflatex)
restricted \write18 enabled.
entering extended mode
|
BSD-2-Clause
|
Tutorial-u0_smallb_Poynting-Cartesian.ipynb
|
KAClough/nrpytutorial
|
T81-558: Applications of Deep Neural Networks**Module 4: Training for Tabular Data*** Instructor: [Jeff Heaton](https://sites.wustl.edu/jeffheaton/), McKelvey School of Engineering, [Washington University in St. Louis](https://engineering.wustl.edu/Programs/Pages/default.aspx)* For more information visit the [class website](https://sites.wustl.edu/jeffheaton/t81-558/). Module 4 Material* **Part 4.1: Encoding a Feature Vector for Keras Deep Learning** [[Video]](https://www.youtube.com/watch?v=Vxz-gfs9nMQ&list=PLjy4p-07OYzulelvJ5KVaT2pDlxivl_BN) [[Notebook]](https://github.com/jeffheaton/t81_558_deep_learning/blob/master/t81_558_class_04_1_feature_encode.ipynb)* Part 4.2: Keras Multiclass Classification for Deep Neural Networks with ROC and AUC [[Video]](https://www.youtube.com/watch?v=-f3bg9dLMks&list=PLjy4p-07OYzulelvJ5KVaT2pDlxivl_BN) [[Notebook]](https://github.com/jeffheaton/t81_558_deep_learning/blob/master/t81_558_class_04_2_multi_class.ipynb)* Part 4.3: Keras Regression for Deep Neural Networks with RMSE [[Video]](https://www.youtube.com/watch?v=wNhBUC6X5-E&list=PLjy4p-07OYzulelvJ5KVaT2pDlxivl_BN) [[Notebook]](https://github.com/jeffheaton/t81_558_deep_learning/blob/master/t81_558_class_04_3_regression.ipynb)* Part 4.4: Backpropagation, Nesterov Momentum, and ADAM Neural Network Training [[Video]](https://www.youtube.com/watch?v=VbDg8aBgpck&list=PLjy4p-07OYzulelvJ5KVaT2pDlxivl_BN) [[Notebook]](https://github.com/jeffheaton/t81_558_deep_learning/blob/master/t81_558_class_04_4_backprop.ipynb)* Part 4.5: Neural Network RMSE and Log Loss Error Calculation from Scratch [[Video]](https://www.youtube.com/watch?v=wmQX1t2PHJc&list=PLjy4p-07OYzulelvJ5KVaT2pDlxivl_BN) [[Notebook]](https://github.com/jeffheaton/t81_558_deep_learning/blob/master/t81_558_class_04_5_rmse_logloss.ipynb) Google CoLab InstructionsThe following code ensures that Google CoLab is running the correct version of TensorFlow.
|
try:
%tensorflow_version 2.x
COLAB = True
print("Note: using Google CoLab")
except:
print("Note: not using Google CoLab")
COLAB = False
|
Note: not using Google CoLab
|
Apache-2.0
|
t81_558_class_04_1_feature_encode.ipynb
|
IlkerCa/t81_558_deep_learning
|
Part 4.1: Encoding a Feature Vector for Keras Deep LearningNeural networks can accept many types of data. We will begin with tabular data, where there are well defined rows and columns. This is the sort of data you would typically see in Microsoft Excel. An example of tabular data is shown below.Neural networks require numeric input. This numeric form is called a feature vector. Each row of training data typically becomes one vector. The individual input neurons each receive one feature (or column) from this vector. In this section, we will see how to encode the following tabular data into a feature vector.
|
import pandas as pd
pd.set_option('display.max_columns', 7)
pd.set_option('display.max_rows', 5)
df = pd.read_csv(
"https://data.heatonresearch.com/data/t81-558/jh-simple-dataset.csv",
na_values=['NA','?'])
pd.set_option('display.max_columns', 9)
pd.set_option('display.max_rows', 5)
display(df)
|
_____no_output_____
|
Apache-2.0
|
t81_558_class_04_1_feature_encode.ipynb
|
IlkerCa/t81_558_deep_learning
|
The following observations can be made from the above data:* The target column is the column that you seek to predict. There are several candidates here. However, we will initially use product. This field specifies what product someone bought.* There is an ID column. This column should not be fed into the neural network as it contains no information useful for prediction.* Many of these fields are numeric and might not require any further processing.* The income column does have some missing values.* There are categorical values: job, area, and product.To begin with, we will convert the job code into dummy variables.
|
pd.set_option('display.max_columns', 7)
pd.set_option('display.max_rows', 5)
dummies = pd.get_dummies(df['job'],prefix="job")
print(dummies.shape)
pd.set_option('display.max_columns', 9)
pd.set_option('display.max_rows', 10)
display(dummies)
|
(2000, 33)
|
Apache-2.0
|
t81_558_class_04_1_feature_encode.ipynb
|
IlkerCa/t81_558_deep_learning
|
Because there are 33 different job codes, there are 33 dummy variables. We also specified a prefix, because the job codes (such as "ax") are not that meaningful by themselves. Something such as "job_ax" also tells us the origin of this field.Next, we must merge these dummies back into the main data frame. We also drop the original "job" field, as it is now represented by the dummies.
|
pd.set_option('display.max_columns', 7)
pd.set_option('display.max_rows', 5)
df = pd.concat([df,dummies],axis=1)
df.drop('job', axis=1, inplace=True)
pd.set_option('display.max_columns', 9)
pd.set_option('display.max_rows', 10)
display(df)
|
_____no_output_____
|
Apache-2.0
|
t81_558_class_04_1_feature_encode.ipynb
|
IlkerCa/t81_558_deep_learning
|
We also introduce dummy variables for the area column.
|
pd.set_option('display.max_columns', 7)
pd.set_option('display.max_rows', 5)
df = pd.concat([df,pd.get_dummies(df['area'],prefix="area")],axis=1)
df.drop('area', axis=1, inplace=True)
pd.set_option('display.max_columns', 9)
pd.set_option('display.max_rows', 10)
display(df)
|
_____no_output_____
|
Apache-2.0
|
t81_558_class_04_1_feature_encode.ipynb
|
IlkerCa/t81_558_deep_learning
|
The last remaining transformation is to fill in missing income values.
|
med = df['income'].median()
df['income'] = df['income'].fillna(med)
|
_____no_output_____
|
Apache-2.0
|
t81_558_class_04_1_feature_encode.ipynb
|
IlkerCa/t81_558_deep_learning
|
There are more advanced ways of filling in missing values, but they require more analysis. The idea would be to see if another field might give a hint as to what the income were. For example, it might be beneficial to calculate a median income for each of the areas or job categories. This is something to keep in mind for the class Kaggle competition.At this point, the Pandas dataframe is ready to be converted to Numpy for neural network training. We need to know a list of the columns that will make up *x* (the predictors or inputs) and *y* (the target). The complete list of columns is:
|
print(list(df.columns))
|
['id', 'income', 'aspect', 'subscriptions', 'dist_healthy', 'save_rate', 'dist_unhealthy', 'age', 'pop_dense', 'retail_dense', 'crime', 'product', 'job_11', 'job_al', 'job_am', 'job_ax', 'job_bf', 'job_by', 'job_cv', 'job_de', 'job_dz', 'job_e2', 'job_f8', 'job_gj', 'job_gv', 'job_kd', 'job_ke', 'job_kl', 'job_kp', 'job_ks', 'job_kw', 'job_mm', 'job_nb', 'job_nn', 'job_ob', 'job_pe', 'job_po', 'job_pq', 'job_pz', 'job_qp', 'job_qw', 'job_rn', 'job_sa', 'job_vv', 'job_zz', 'area_a', 'area_b', 'area_c', 'area_d']
|
Apache-2.0
|
t81_558_class_04_1_feature_encode.ipynb
|
IlkerCa/t81_558_deep_learning
|
This includes both the target and predictors. We need a list with the target removed. We also remove **id** because it is not useful for prediction.
|
x_columns = df.columns.drop('product').drop('id')
print(list(x_columns))
|
['income', 'aspect', 'subscriptions', 'dist_healthy', 'save_rate', 'dist_unhealthy', 'age', 'pop_dense', 'retail_dense', 'crime', 'job_11', 'job_al', 'job_am', 'job_ax', 'job_bf', 'job_by', 'job_cv', 'job_de', 'job_dz', 'job_e2', 'job_f8', 'job_gj', 'job_gv', 'job_kd', 'job_ke', 'job_kl', 'job_kp', 'job_ks', 'job_kw', 'job_mm', 'job_nb', 'job_nn', 'job_ob', 'job_pe', 'job_po', 'job_pq', 'job_pz', 'job_qp', 'job_qw', 'job_rn', 'job_sa', 'job_vv', 'job_zz', 'area_a', 'area_b', 'area_c', 'area_d']
|
Apache-2.0
|
t81_558_class_04_1_feature_encode.ipynb
|
IlkerCa/t81_558_deep_learning
|
Generate X and Y for a Classification Neural Network We can now generate *x* and *y*. Note, this is how we generate y for a classification problem. Regression would not use dummies and would simply encode the numeric value of the target.
|
# Convert to numpy - Classification
x_columns = df.columns.drop('product').drop('id')
x = df[x_columns].values
dummies = pd.get_dummies(df['product']) # Classification
products = dummies.columns
y = dummies.values
|
_____no_output_____
|
Apache-2.0
|
t81_558_class_04_1_feature_encode.ipynb
|
IlkerCa/t81_558_deep_learning
|
We can display the *x* and *y* matrices.
|
print(x)
print(y)
|
[[5.08760000e+04 1.31000000e+01 1.00000000e+00 ... 0.00000000e+00
1.00000000e+00 0.00000000e+00]
[6.03690000e+04 1.86250000e+01 2.00000000e+00 ... 0.00000000e+00
1.00000000e+00 0.00000000e+00]
[5.51260000e+04 3.47666667e+01 1.00000000e+00 ... 0.00000000e+00
1.00000000e+00 0.00000000e+00]
...
[2.85950000e+04 3.94250000e+01 3.00000000e+00 ... 0.00000000e+00
0.00000000e+00 1.00000000e+00]
[6.79490000e+04 5.73333333e+00 0.00000000e+00 ... 0.00000000e+00
1.00000000e+00 0.00000000e+00]
[6.14670000e+04 1.68916667e+01 0.00000000e+00 ... 0.00000000e+00
1.00000000e+00 0.00000000e+00]]
[[0 1 0 ... 0 0 0]
[0 0 1 ... 0 0 0]
[0 1 0 ... 0 0 0]
...
[0 0 0 ... 0 1 0]
[0 0 1 ... 0 0 0]
[0 0 1 ... 0 0 0]]
|
Apache-2.0
|
t81_558_class_04_1_feature_encode.ipynb
|
IlkerCa/t81_558_deep_learning
|
The x and y values are now ready for a neural network. Make sure that you construct the neural network for a classification problem. Specifically,* Classification neural networks have an output neuron count equal to the number of classes.* Classification neural networks should use **categorical_crossentropy** and a **softmax** activation function on the output layer. Generate X and Y for a Regression Neural NetworkFor a regression neural network, the *x* values are generated the same. However, *y* does not use dummies. Make sure to replace **income** with your actual target.
|
y = df['income'].values
|
_____no_output_____
|
Apache-2.0
|
t81_558_class_04_1_feature_encode.ipynb
|
IlkerCa/t81_558_deep_learning
|
**This notebook is an exercise in the [Intermediate Machine Learning](https://www.kaggle.com/learn/intermediate-machine-learning) course. You can reference the tutorial at [this link](https://www.kaggle.com/alexisbcook/categorical-variables).**--- By encoding **categorical variables**, you'll obtain your best results thus far! SetupThe questions below will give you feedback on your work. Run the following cell to set up the feedback system.
|
# Set up code checking
import os
if not os.path.exists("../input/train.csv"):
os.symlink("../input/home-data-for-ml-course/train.csv", "../input/train.csv")
os.symlink("../input/home-data-for-ml-course/test.csv", "../input/test.csv")
from learntools.core import binder
binder.bind(globals())
from learntools.ml_intermediate.ex3 import *
print("Setup Complete")
|
Setup Complete
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
In this exercise, you will work with data from the [Housing Prices Competition for Kaggle Learn Users](https://www.kaggle.com/c/home-data-for-ml-course). Run the next code cell without changes to load the training and validation sets in `X_train`, `X_valid`, `y_train`, and `y_valid`. The test set is loaded in `X_test`.
|
import pandas as pd
from sklearn.model_selection import train_test_split
# Read the data
X = pd.read_csv('../input/train.csv', index_col='Id')
X_test = pd.read_csv('../input/test.csv', index_col='Id')
# Remove rows with missing target, separate target from predictors
X.dropna(axis=0, subset=['SalePrice'], inplace=True)
y = X.SalePrice
X.drop(['SalePrice'], axis=1, inplace=True)
# To keep things simple, we'll drop columns with missing values
cols_with_missing = [col for col in X.columns if X[col].isnull().any()]
X.drop(cols_with_missing, axis=1, inplace=True)
X_test.drop(cols_with_missing, axis=1, inplace=True)
# Break off validation set from training data
X_train, X_valid, y_train, y_valid = train_test_split(X, y,
train_size=0.8, test_size=0.2,
random_state=0)
|
_____no_output_____
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Use the next code cell to print the first five rows of the data.
|
X_train.head()
|
_____no_output_____
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Notice that the dataset contains both numerical and categorical variables. You'll need to encode the categorical data before training a model.To compare different models, you'll use the same `score_dataset()` function from the tutorial. This function reports the [mean absolute error](https://en.wikipedia.org/wiki/Mean_absolute_error) (MAE) from a random forest model.
|
from sklearn.ensemble import RandomForestRegressor
from sklearn.metrics import mean_absolute_error
# function for comparing different approaches
def score_dataset(X_train, X_valid, y_train, y_valid):
model = RandomForestRegressor(n_estimators=100, random_state=0)
model.fit(X_train, y_train)
preds = model.predict(X_valid)
return mean_absolute_error(y_valid, preds)
|
_____no_output_____
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Step 1: Drop columns with categorical dataYou'll get started with the most straightforward approach. Use the code cell below to preprocess the data in `X_train` and `X_valid` to remove columns with categorical data. Set the preprocessed DataFrames to `drop_X_train` and `drop_X_valid`, respectively.
|
# Fill in the lines below: drop columns in training and validation data
drop_X_train = X_train.select_dtypes(exclude=['object'])
drop_X_valid = X_valid.select_dtypes(exclude=['object'])
# Check your answers
step_1.check()
# Lines below will give you a hint or solution code
#step_1.hint()
#step_1.solution()
|
_____no_output_____
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Run the next code cell to get the MAE for this approach.
|
print("MAE from Approach 1 (Drop categorical variables):")
print(score_dataset(drop_X_train, drop_X_valid, y_train, y_valid))
|
MAE from Approach 1 (Drop categorical variables):
17837.82570776256
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Before jumping into label encoding, we'll investigate the dataset. Specifically, we'll look at the `'Condition2'` column. The code cell below prints the unique entries in both the training and validation sets.
|
print("Unique values in 'Condition2' column in training data:", X_train['Condition2'].unique())
print("\nUnique values in 'Condition2' column in validation data:", X_valid['Condition2'].unique())
|
Unique values in 'Condition2' column in training data: ['Norm' 'PosA' 'Feedr' 'PosN' 'Artery' 'RRAe']
Unique values in 'Condition2' column in validation data: ['Norm' 'RRAn' 'RRNn' 'Artery' 'Feedr' 'PosN']
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Step 2: Label encoding Part AIf you now write code to: - fit a label encoder to the training data, and then - use it to transform both the training and validation data, you'll get an error. Can you see why this is the case? (_You'll need to use the above output to answer this question._)
|
# Check your answer (Run this code cell to receive credit!)
step_2.a.check()
#step_2.a.hint()
|
_____no_output_____
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
This is a common problem that you'll encounter with real-world data, and there are many approaches to fixing this issue. For instance, you can write a custom label encoder to deal with new categories. The simplest approach, however, is to drop the problematic categorical columns. Run the code cell below to save the problematic columns to a Python list `bad_label_cols`. Likewise, columns that can be safely label encoded are stored in `good_label_cols`.
|
# All categorical columns
object_cols = [col for col in X_train.columns if X_train[col].dtype == "object"]
# Columns that can be safely label encoded
good_label_cols = [col for col in object_cols if
set(X_train[col]) == set(X_valid[col])]
# Problematic columns that will be dropped from the dataset
bad_label_cols = list(set(object_cols)-set(good_label_cols))
print('Categorical columns that will be label encoded:', good_label_cols)
print('\nCategorical columns that will be dropped from the dataset:', bad_label_cols)
|
Categorical columns that will be label encoded: ['MSZoning', 'Street', 'LotShape', 'LandContour', 'LotConfig', 'BldgType', 'HouseStyle', 'ExterQual', 'CentralAir', 'KitchenQual', 'PavedDrive', 'SaleCondition']
Categorical columns that will be dropped from the dataset: ['Neighborhood', 'LandSlope', 'Condition1', 'Heating', 'Foundation', 'RoofMatl', 'Condition2', 'RoofStyle', 'ExterCond', 'Exterior1st', 'Utilities', 'Functional', 'HeatingQC', 'SaleType', 'Exterior2nd']
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Part BUse the next code cell to label encode the data in `X_train` and `X_valid`. Set the preprocessed DataFrames to `label_X_train` and `label_X_valid`, respectively. - We have provided code below to drop the categorical columns in `bad_label_cols` from the dataset. - You should label encode the categorical columns in `good_label_cols`.
|
from sklearn.preprocessing import LabelEncoder
# Drop categorical columns that will not be encoded
label_X_train = X_train.drop(bad_label_cols, axis=1)
label_X_valid = X_valid.drop(bad_label_cols, axis=1)
# Apply label encoder
label_encoder = LabelEncoder()
for col in good_label_cols:
label_X_train[col] = label_encoder.fit_transform(label_X_train[col])
label_X_valid[col] = label_encoder.transform(label_X_valid[col])
# Check your answer
step_2.b.check()
# Lines below will give you a hint or solution code
#step_2.b.hint()
#step_2.b.solution()
|
_____no_output_____
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Run the next code cell to get the MAE for this approach.
|
print("MAE from Approach 2 (Label Encoding):")
print(score_dataset(label_X_train, label_X_valid, y_train, y_valid))
|
MAE from Approach 2 (Label Encoding):
17575.291883561644
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
So far, you've tried two different approaches to dealing with categorical variables. And, you've seen that encoding categorical data yields better results than removing columns from the dataset.Soon, you'll try one-hot encoding. Before then, there's one additional topic we need to cover. Begin by running the next code cell without changes.
|
# Get number of unique entries in each column with categorical data
object_nunique = list(map(lambda col: X_train[col].nunique(), object_cols))
d = dict(zip(object_cols, object_nunique))
# Print number of unique entries by column, in ascending order
sorted(d.items(), key=lambda x: x[1])
|
_____no_output_____
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Step 3: Investigating cardinality Part AThe output above shows, for each column with categorical data, the number of unique values in the column. For instance, the `'Street'` column in the training data has two unique values: `'Grvl'` and `'Pave'`, corresponding to a gravel road and a paved road, respectively.We refer to the number of unique entries of a categorical variable as the **cardinality** of that categorical variable. For instance, the `'Street'` variable has cardinality 2.Use the output above to answer the questions below.
|
# Fill in the line below: How many categorical variables in the training data
# have cardinality greater than 10?
high_cardinality_numcols = 3
# Fill in the line below: How many columns are needed to one-hot encode the
# 'Neighborhood' variable in the training data?
num_cols_neighborhood = 25
# Check your answers
step_3.a.check()
# Lines below will give you a hint or solution code
#step_3.a.hint()
#step_3.a.solution()
|
_____no_output_____
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Part BFor large datasets with many rows, one-hot encoding can greatly expand the size of the dataset. For this reason, we typically will only one-hot encode columns with relatively low cardinality. Then, high cardinality columns can either be dropped from the dataset, or we can use label encoding.As an example, consider a dataset with 10,000 rows, and containing one categorical column with 100 unique entries. - If this column is replaced with the corresponding one-hot encoding, how many entries are added to the dataset? - If we instead replace the column with the label encoding, how many entries are added? Use your answers to fill in the lines below.
|
# Fill in the line below: How many entries are added to the dataset by
# replacing the column with a one-hot encoding?
OH_entries_added = 1e4*100 - 1e4
# Fill in the line below: How many entries are added to the dataset by
# replacing the column with a label encoding?
label_entries_added = 0
# Check your answers
step_3.b.check()
# Lines below will give you a hint or solution code
#step_3.b.hint()
#step_3.b.solution()
|
_____no_output_____
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Next, you'll experiment with one-hot encoding. But, instead of encoding all of the categorical variables in the dataset, you'll only create a one-hot encoding for columns with cardinality less than 10.Run the code cell below without changes to set `low_cardinality_cols` to a Python list containing the columns that will be one-hot encoded. Likewise, `high_cardinality_cols` contains a list of categorical columns that will be dropped from the dataset.
|
# Columns that will be one-hot encoded
low_cardinality_cols = [col for col in object_cols if X_train[col].nunique() < 10]
# Columns that will be dropped from the dataset
high_cardinality_cols = list(set(object_cols)-set(low_cardinality_cols))
print('Categorical columns that will be one-hot encoded:', low_cardinality_cols)
print('\nCategorical columns that will be dropped from the dataset:', high_cardinality_cols)
|
Categorical columns that will be one-hot encoded: ['MSZoning', 'Street', 'LotShape', 'LandContour', 'Utilities', 'LotConfig', 'LandSlope', 'Condition1', 'Condition2', 'BldgType', 'HouseStyle', 'RoofStyle', 'RoofMatl', 'ExterQual', 'ExterCond', 'Foundation', 'Heating', 'HeatingQC', 'CentralAir', 'KitchenQual', 'Functional', 'PavedDrive', 'SaleType', 'SaleCondition']
Categorical columns that will be dropped from the dataset: ['Neighborhood', 'Exterior2nd', 'Exterior1st']
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Step 4: One-hot encodingUse the next code cell to one-hot encode the data in `X_train` and `X_valid`. Set the preprocessed DataFrames to `OH_X_train` and `OH_X_valid`, respectively. - The full list of categorical columns in the dataset can be found in the Python list `object_cols`.- You should only one-hot encode the categorical columns in `low_cardinality_cols`. All other categorical columns should be dropped from the dataset.
|
from sklearn.preprocessing import OneHotEncoder
# Use as many lines of code as you need!
OH_encoder = OneHotEncoder(handle_unknown='ignore', sparse=False)
OH_cols_train = pd.DataFrame(OH_encoder.fit_transform(X_train[low_cardinality_cols]))
OH_cols_valid = pd.DataFrame(OH_encoder.transform(X_valid[low_cardinality_cols]))
# One-hot encoding removed index; put it back
OH_cols_train.index = X_train.index
OH_cols_valid.index = X_valid.index
# Remove categorical columns (will replace with one-hot encoding)
num_X_train = X_train.drop(object_cols, axis=1)
num_X_valid = X_valid.drop(object_cols, axis=1)
# Add one-hot encoded columns to numerical features
OH_X_train = pd.concat([num_X_train, OH_cols_train], axis=1)
OH_X_valid = pd.concat([num_X_valid, OH_cols_valid], axis=1)
# Check your answer
step_4.check()
# Lines below will give you a hint or solution code
#step_4.hint()
#step_4.solution()
|
_____no_output_____
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Run the next code cell to get the MAE for this approach.
|
print("MAE from Approach 3 (One-Hot Encoding):")
print(score_dataset(OH_X_train, OH_X_valid, y_train, y_valid))
|
_____no_output_____
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Generate test predictions and submit your resultsAfter you complete Step 4, if you'd like to use what you've learned to submit your results to the leaderboard, you'll need to preprocess the test data before generating predictions.**This step is completely optional, and you do not need to submit results to the leaderboard to successfully complete the exercise.**Check out the previous exercise if you need help with remembering how to [join the competition](https://www.kaggle.com/c/home-data-for-ml-course) or save your results to CSV. Once you have generated a file with your results, follow the instructions below:1. Begin by clicking on the blue **Save Version** button in the top right corner of the window. This will generate a pop-up window. 2. Ensure that the **Save and Run All** option is selected, and then click on the blue **Save** button.3. This generates a window in the bottom left corner of the notebook. After it has finished running, click on the number to the right of the **Save Version** button. This pulls up a list of versions on the right of the screen. Click on the ellipsis **(...)** to the right of the most recent version, and select **Open in Viewer**. This brings you into view mode of the same page. You will need to scroll down to get back to these instructions.4. Click on the **Output** tab on the right of the screen. Then, click on the file you would like to submit, and click on the blue **Submit** button to submit your results to the leaderboard.You have now successfully submitted to the competition!If you want to keep working to improve your performance, select the blue **Edit** button in the top right of the screen. Then you can change your code and repeat the process. There's a lot of room to improve, and you will climb up the leaderboard as you work.
|
# (Optional) Your code here
|
_____no_output_____
|
Apache-2.0
|
pre_exercises/Intermediate_ML/exercise-categorical-variables.ipynb
|
krishnaaxo/Spotify_Skip_Action_Prediction
|
Copyright 2018 The TensorFlow Authors.Licensed under the Apache License, Version 2.0 (the "License");
|
#@title Licensed under the Apache License, Version 2.0 (the "License"); { display-mode: "form" }
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
|
_____no_output_____
|
Apache-2.0
|
site/en/r2/guide/_tpu.ipynb
|
christophmeyer/docs
|
View on TensorFlow.org Run in Google Colab View source on GitHub Using TPUsTensor Processing Units (TPUs) are Google's specialized ASICs designed to dramatically accelerate machine learning workloads. They are available on Google Colab, the TensorFlow Research Cloud and Google Compute Engine. In this notebook, you can try training a convolutional neural network against the Fashion MNIST dataset on Cloud TPUs using tf.keras and Distribution Strategy. Learning ObjectivesIn this Colab, you will learn how to:* Write a standard 4-layer conv-net with drop-out and batch normalization in Keras.* Use TPUs and Distribution Strategy to train the model.* Run a prediction to see how well the model can predict fashion categories and output the result. InstructionsTo use TPUs in Colab:1. On the main menu, click Runtime and select **Change runtime type**. Set "TPU" as the hardware accelerator.1. Click Runtime again and select **Runtime > Run All**. You can also run the cells manually with Shift-ENTER. Data, Model, and Training Download the DataBegin by downloading the fashion MNIST dataset using `tf.keras.datasets`, as shown below. We will also need to convert the data to `float32` format, as the data types supported by TPUs are limited right now.TPUs currently do not support Eager Execution, so we disable that with `disable_eager_execution()`.
|
from __future__ import absolute_import, division, print_function, unicode_literals
import numpy as np
from __future__ import absolute_import, division, print_function
!pip install tensorflow-gpu==2.0.0-beta1
import tensorflow as tf
tf.compat.v1.disable_eager_execution()
import numpy as np
(x_train, y_train), (x_test, y_test) = tf.keras.datasets.fashion_mnist.load_data()
# add empty color dimension
x_train = np.expand_dims(x_train, -1)
x_test = np.expand_dims(x_test, -1)
# convert types to float32
x_train = x_train.astype(np.float32)
x_test = x_test.astype(np.float32)
y_train = y_train.astype(np.float32)
y_test = y_test.astype(np.float32)
|
_____no_output_____
|
Apache-2.0
|
site/en/r2/guide/_tpu.ipynb
|
christophmeyer/docs
|
Initialize TPUStrategyWe first initialize the TPUStrategy object before creating the model, so that Keras knows that we are creating a model for TPUs. To do this, we are first creating a TPUClusterResolver using the IP address of the TPU, and then creating a TPUStrategy object from the Cluster Resolver.
|
import os
resolver = tf.distribute.cluster_resolver.TPUClusterResolver()
tf.tpu.experimental.initialize_tpu_system(resolver)
strategy = tf.distribute.experimental.TPUStrategy(resolver)
|
_____no_output_____
|
Apache-2.0
|
site/en/r2/guide/_tpu.ipynb
|
christophmeyer/docs
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.