kernel_id
int64 24.2k
23.3M
| prompt
stringlengths 8
1.85M
| completetion
stringlengths 1
182k
| comp_name
stringlengths 5
57
|
|---|---|---|---|
11,614,110 | test['Title'] = test['Title'].replace(['Lady.', 'Capt.', 'Col.',
'Don.', 'Dr.', 'Major.', 'Rev.', 'Jonkheer.', 'Dona.'], 'Rare.')
test['Title'] = test['Title'].replace(['Countess.', 'Lady.', 'Sir.'], 'Royal.')
test['Title'] = test['Title'].replace('Mlle.', 'Miss')
test['Title'] = test['Title'].replace('Ms.', 'Miss.')
test['Title'] = test['Title'].replace('Mme.', 'Mrs.')
<categorify> | train.id.nunique() | Conway's Reverse Game of Life 2020 |
11,614,110 | title_mapping = {"Mr.": 1, "Miss.": 2, "Mrs.": 3, "Master.": 4, "Royal.": 5, "Rare.": 6}
train['Title'] = train['Title'].map(title_mapping)
train['Title'] = train['Title'].fillna(0)
train.head()<categorify> | train.delta.value_counts() | Conway's Reverse Game of Life 2020 |
11,614,110 | title_mapping = {"Mr.": 1, "Miss.": 2, "Mrs.": 3, "Master.": 4, "Royal.": 5, "Rare.": 6}
test['Title'] = test['Title'].map(title_mapping)
test['Title'] = test['Title'].fillna(0)
<drop_column> | train.start_0.value_counts() | Conway's Reverse Game of Life 2020 |
11,614,110 | test = test.drop(['Ticket'], axis = 1)
test = test.drop(['Name'], axis = 1)
test = test.drop(['Parch'], axis = 1)
test = test.drop(['Fare','SibSp'], axis = 1)
train.drop(['Name', 'Ticket'], axis = 1, inplace = True)
train = train.drop(['Parch'], axis = 1)
train = train.drop(['Fare','SibSp'], axis = 1 )<split> | train.stop_0.value_counts() | Conway's Reverse Game of Life 2020 |
11,614,110 | X_train, X_test, y_train, y_test = train_test_split(train.drop(['Survived','PassengerId'], axis=1),
train['Survived'], test_size = 0.2,
random_state = 0 )<train_model> | sample_start_1 = train.loc[1, train.columns.str.startswith('start')]
sample_stop_1 = train.loc[1, train.columns.str.startswith('stop')] | Conway's Reverse Game of Life 2020 |
11,614,110 | logisticRegression = LogisticRegression(max_iter = 30000)
logisticRegression.fit(X_train, y_train)
<predict_on_test> | sample_start_1 = np.asarray(sample_start_1 ).reshape(25, 25)
sample_stop_1 = np.asarray(sample_stop_1 ).reshape(25, 25 ) | Conway's Reverse Game of Life 2020 |
11,614,110 | predictions = logisticRegression.predict(X_test)
acc_LOG = round(accuracy_score(predictions, y_test)* 100, 2)
print(acc_LOG)
print(predictions)
<compute_test_metric> | train.sum() [1:].sort_values() | Conway's Reverse Game of Life 2020 |
11,614,110 | print(confusion_matrix(y_test, predictions))<compute_test_metric> | gs=train[:1][[col for col in train.columns if 'start' in col]].values.reshape(-1,1 ) | Conway's Reverse Game of Life 2020 |
11,614,110 | accuracy=(( 88+50)/(88+50+22+19))
print('accuracy is: ',(round(accuracy, 2)*100))<train_model> | %%cython
c
@cython.cdivision(True)
@cython.boundscheck(False)
@cython.nonecheck(False)
@cython.wraparound(False)
cdef int calc_neighs(unsigned char[:, :] field, int i, int j, int n, int k):
cdef:
int neighs = 0;
int i_min = i - 1;
int i_pl = i + 1;
int j_min = j - 1;
int j_pl = j + 1;
neighs = 0
if i_min >= 0:
if j_min >= 0:
neighs += field[i_min, j_min]
neighs += field[i_min, j]
if j_pl < k:
neighs += field[i_min, j_pl]
if j_min >= 0:
neighs += field[i, j_min]
if j_pl < k:
neighs += field[i, j_pl]
if i_pl < n:
if j_min >= 0:
neighs += field[i_pl, j_min]
neighs += field[i_pl, j]
if j_pl < k:
neighs += field[i_pl, j_pl]
return neighs
@cython.cdivision(True)
@cython.boundscheck(False)
@cython.nonecheck(False)
@cython.wraparound(False)
cpdef make_move(unsigned char[:, :] field, int moves):
cdef:
int _, i, j, neighs;
int n, k;
int switch = 0;
unsigned char[:, :] cur_field;
unsigned char[:, :] next_field;
cur_field = np.copy(field)
next_field = np.zeros_like(field, 'uint8')
n = field.shape[0]
k = field.shape[1]
for _ in range(moves):
if switch == 0:
for i in range(n):
for j in range(k):
neighs = calc_neighs(cur_field, i, j, n, k)
if cur_field[i, j] and neighs == 2:
next_field[i, j] = 1
elif neighs == 3:
next_field[i, j] = 1
else:
next_field[i, j] = 0
else:
for i in range(n):
for j in range(k):
neighs = calc_neighs(next_field, i, j, n, k)
if next_field[i, j] and neighs == 2:
cur_field[i, j] = 1
elif neighs == 3:
cur_field[i, j] = 1
else:
cur_field[i, j] = 0
switch =(switch + 1)% 2
return np.array(next_field if switch else cur_field ) | Conway's Reverse Game of Life 2020 |
11,614,110 | randomforest = RandomForestClassifier(random_state = 5, criterion = 'gini', max_depth = 10, max_features = 'auto', n_estimators = 500)
randomforest.fit(X_train, y_train)
pred = randomforest.predict(X_test)
acc_randomforest = round(accuracy_score(pred, y_test)* 100, 2)
print(acc_randomforest)
<train_model> | NROW, NCOL = 25, 25
def generate_samples(delta=1, n=32):
batch = np.split(np.random.binomial(1, 0.5,(NROW * n, NCOL)).astype('uint8'), n)
Yy = [life.make_move(state, 5)for state in batch]
Xx = [life.make_move(state, 1)for state in Yy]
Y = np.array([y.ravel() for y in Yy])
X = np.array([x.ravel() for x in Xx])
return X, Y
def data_generator(delta=1, batch_size=32, ravel=True):
while True:
batch = np.split(np.random.binomial(1, 0.5,(NROW * batch_size, NCOL)).astype('uint8'),
batch_size)
Yy = [make_move(state, 5)for state in batch]
Xx = [make_move(state, delta)for state in Yy]
if ravel:
Y = np.array([y.ravel() for y in Yy])
X = np.array([x.ravel() for x in Xx])
yield X, Y
else:
yield np.array(Xx)[:,:, :, np.newaxis], np.array(Yy)[:, :, :, np.newaxis] | Conway's Reverse Game of Life 2020 |
11,614,110 | gbk = GradientBoostingClassifier()
gbk.fit(X_train, y_train)
pred = gbk.predict(X_test)
acc_gbk = round(accuracy_score(pred, y_test)* 100, 2)
print(acc_gbk )<create_dataframe> | def create_model(n_hidden_convs=2, n_hidden_filters=128, kernel_size=5):
nn = Sequential()
nn.add(Conv2D(n_hidden_filters, kernel_size, padding='same', activation='relu',
input_shape=(25, 25, 1)))
nn.add(BatchNormalization())
for i in range(n_hidden_convs):
nn.add(Conv2D(n_hidden_filters, kernel_size, padding='same', activation='relu'))
nn.add(BatchNormalization())
nn.add(Conv2D(1, kernel_size, padding='same', activation='sigmoid'))
nn.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])
return nn | Conway's Reverse Game of Life 2020 |
11,614,110 | see={'TECHNIQUE':['RANDOM FOREST','LOGISTIC REGRESSION','GRADIENT BOOSTING'],'ACCURACY':[acc_randomforest,acc_LOG,acc_gbk]}
mod=pd.DataFrame(see)
mod<save_to_csv> | models = []
for delta in range(1, 6):
model = create_model(n_hidden_convs=6, n_hidden_filters=256)
es = EarlyStopping(monitor='loss', patience=9, min_delta=0.001)
model.fit_generator(data_generator(delta=delta, ravel=False),
steps_per_epoch=500, epochs=50, verbose=1, callbacks=[es])
models.append(model ) | Conway's Reverse Game of Life 2020 |
11,614,110 | ids = test['PassengerId']
predictions = randomforest.predict(test.drop('PassengerId', axis=1))
output = pd.DataFrame({ 'PassengerId' : ids, 'Survived': predictions })
output.to_csv('submission.csv', index=False )<set_options> | test = pd.read_csv('.. /input/conways-reverse-game-of-life-2020/test.csv', index_col='id' ) | Conway's Reverse Game of Life 2020 |
11,614,110 | np.random.seed(1234 )<load_from_csv> | submit_df = pd.DataFrame(index=test.index, columns=['start_' + str(i)for i in range(625)] ) | Conway's Reverse Game of Life 2020 |
11,614,110 | dataset = pd.read_csv('.. /input/titanic/train.csv')
test = pd.read_csv('.. /input/titanic/test.csv')
y = dataset['Survived'].values
dataset.head()<data_type_conversions> | for delta in range(1, 6):
mod = models[delta-1]
delta_df = test[test.delta == delta].iloc[:, 1:].values.reshape(-1, 25, 25, 1)
submit_df[test.delta == delta] = mod.predict(delta_df ).reshape(-1, 625 ).round(0 ).astype('uint8' ) | Conway's Reverse Game of Life 2020 |
11,614,110 | <feature_engineering><EOS> | submit_df.to_csv('submission.csv' ) | Conway's Reverse Game of Life 2020 |
14,018,577 | <SOS> metric: custom netric Kaggle data source: nfl-impact-detection<train_on_grid> | !tar xfz.. /input/nfl-lib/pkgs.tgz
| NFL 1st and Future - Impact Detection |
14,018,577 | model = KNeighborsClassifier()
hyperparameters = {
"n_neighbors" : range(1,20,2),
'weights' : ['uniform', 'distance'],
'p' : [1, 2]
}
grid = GridSearchCV(model, param_grid=hyperparameters, cv=10)
grid.fit(X, y)
best_params = grid.best_params_
best_score = grid.best_score_
knn = grid.best_estimator_
y_pred = knn.predict(X_test)
print(grid.best_params_)
print(grid.best_estimator_)
print(grid.best_score_ )<train_model> | warnings.filterwarnings("ignore" ) | NFL 1st and Future - Impact Detection |
14,018,577 | perm = PermutationImportance(knn, random_state=1 ).fit(X_test, y_test)
eli5.show_weights(perm, feature_names = X_test.columns.tolist() )<compute_train_metric> | d = pd.read_csv('.. /input/nfl-impact-detection/test_player_tracking.csv')
IS_PRIVATE = d.shape !=(19269, 12)
IS_PRIVATE = True
| NFL 1st and Future - Impact Detection |
14,018,577 | dt = DecisionTreeClassifier()
dt.fit(X_train, y_train)
y_pred = dt.predict(X_test)
accuracy_score(y_pred, y_test)
<save_to_csv> | DATA_ROOT_PATH = 'test_images' | NFL 1st and Future - Impact Detection |
14,018,577 | dot_data = tree.export_graphviz(dt, out_file=None)
graph = graphviz.Source(dot_data)
graph.render("iris")
dot_data = tree.export_graphviz(dt, out_file=None,
feature_names=X_train.columns,
class_names=['Survived', 'Not Survived'],
filled=True, rounded=True,
special_characters=True)
graph = graphviz.Source(dot_data)
graph<train_on_grid> | import argparse
from utils.datasets import *
from utils.general import non_max_suppression
from utils.general import *
from utils import torch_utils
from utils.plots import plot_one_box | NFL 1st and Future - Impact Detection |
14,018,577 | hyperparameters = {"criterion": ["entropy", "gini"],
"max_depth": [3, 5, 7, 10],
"max_features": ["log2", "sqrt", 'auto'],
'min_samples_leaf' : [2, 3, 4, 5],
'min_samples_split' : [2, 3, 4, 5]
}
grid = GridSearchCV(dt, param_grid=hyperparameters, cv=10)
grid.fit(X, y)
best_params = grid.best_params_
best_score = grid.best_score_
dt = grid.best_estimator_
y_pred = dt.predict(X_test)
print(grid.best_params_)
print(grid.best_score_)
plt.figure(figsize=(3,3))
sns.heatmap(confusion_matrix(y_pred, y_test), annot=True, cbar=False, fmt='1d', cmap='Blues')
perm = PermutationImportance(dt, random_state=1 ).fit(X_test, y_test)
eli5.show_weights(perm, feature_names = X_test.columns.tolist() )<train_on_grid> | names = ['0','1','2']
def detect(save_img=False):
weights, imgsz = opt.weights,opt.img_size
source = 'test_images/'
device = torch_utils.select_device(opt.device)
half = False
model = torch.load(weights, map_location=device)['model'].to(device ).float().eval()
dataset = LoadImages(source, img_size=opt.img_size)
t0 = time.time()
img = torch.zeros(( 1, 3, imgsz, imgsz), device=device)
all_path=[]
all_bboxex =[]
all_score =[]
all_c = []
for path, img, im0s, vid_cap in tqdm(dataset):
print(im0s.shape)
img = torch.from_numpy(img ).to(device)
img = img.float()
img /= 255.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
t1 = torch_utils.time_synchronized()
bboxes_2 = []
score_2 = []
c_2 = []
if True:
pred = model(img, augment=opt.augment)[0]
pred = non_max_suppression(pred, opt.conf_thres, opt.iou_thres, classes=None, agnostic=False)
t2 = torch_utils.time_synchronized()
flag=False
bboxes = []
score = []
cc = []
for i, det in enumerate(pred):
p, s, im0 = path, '', im0s
gn = torch.tensor(im0.shape)[[1, 0, 1, 0]]
if det is not None and len(det):
det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape ).round()
for c in det[:, -1].unique() :
n =(det[:, -1] == c ).sum()
for *xyxy, conf, cls in det:
xywh = torch.tensor(xyxy ).view(-1 ).numpy()
bboxes.append(xywh)
score.append(conf.cpu().numpy())
cc.append(cls.cpu().numpy())
bboxes_2.append(bboxes)
score_2.append(score)
c_2.append(cc)
all_path.append(path)
all_score.append(score_2)
all_bboxex.append(bboxes_2)
all_c.append(c_2)
return all_path,all_score,all_bboxex,all_c
if __name__ == '__main__':
class opt:
weights = ".. /input/yolov5temp1280/best(2 ).pt"
img_size = 1280
conf_thres = 0.1
iou_thres = 0.3
augment = False
device = '0'
classes=None
agnostic_nms = True
opt.img_size = check_img_size(opt.img_size)
print(opt)
with torch.no_grad() :
res = detect() | NFL 1st and Future - Impact Detection |
14,018,577 | model = RandomForestClassifier()
hyperparameters = {"criterion": ["entropy", "gini"],
"max_depth": [5, 10],
"max_features": ["log2", "sqrt"],
'min_samples_leaf' : [2, 3, 4, 5],
'min_samples_split' : [2, 3, 4, 5],
"n_estimators": [6, 9]
}
grid = GridSearchCV(model, param_grid=hyperparameters, cv=10)
grid.fit(X, y)
best_params = grid.best_params_
best_score = grid.best_score_
rf = grid.best_estimator_
y_pred = rf.predict(X_test)
print(grid.best_params_)
print(grid.best_score_)
plt.figure(figsize=(3,3))
sns.heatmap(confusion_matrix(y_pred, y_test), annot=True, cbar=False, fmt='1d', cmap='Blues')
perm = PermutationImportance(rf, random_state=1 ).fit(X_test, y_test)
eli5.show_weights(perm, feature_names = X_test.columns.tolist() )<save_to_csv> | %matplotlib inline
all_path,all_score,all_bboxex,all_labels = res
results =[]
results_boxes =[]
results_scores = []
results_labels = []
result_image_ids =[]
for row in range(len(all_path)) :
image_id = all_path[row].split("/")[-1]
boxes = np.array(all_bboxex[row])[0]
scores = np.array(all_score[row])[0]
labels = np.array(all_labels[row])[0]
if len(boxes)==0:
continue
boxes[:, 2] = boxes[:, 2] - boxes[:, 0]
boxes[:, 3] = boxes[:, 3] - boxes[:, 1]
boxes = boxes.astype(int)
if len(boxes)>0:
result_image_ids += [image_id]*len(boxes)
results_boxes.append(boxes)
results_scores.append(scores)
results_labels.append(labels)
idx = row
idx = 0
size = 300
idx =-1
font = cv2.FONT_HERSHEY_SIMPLEX
image = image = cv2.imread(all_path[idx], cv2.IMREAD_COLOR)
fontScale = 1
boxes = results_boxes[idx]
scores = results_scores[idx]
label = results_labels[idx]
color =(255, 0, 0)
thickness = 2
for b,s,l in zip(boxes,scores,label):
color =(255,0,0)if int(l)==2 else(0,0,255)
image = cv2.rectangle(image,(b[0],b[1]),(b[0]+b[2],b[1]+b[3]), color, 1)
image = cv2.putText(image, '{:.2}'.format(s),(int(b[0]),int(b[1])) , font, fontScale, color, thickness, cv2.LINE_AA)
plt.figure(figsize=[20,20])
plt.imshow(image[:,:,::-1])
plt.show() | NFL 1st and Future - Impact Detection |
14,018,577 | holdout_ids = test["PassengerId"]
holdout_features = test[features]
holdout_predictions = lr.predict(holdout_features)
submission = pd.DataFrame({"PassengerId": holdout_ids,
"Survived": holdout_predictions})
print(submission.head())
submission.to_csv("submission.csv",index=False )<load_from_csv> | box_df = pd.DataFrame(np.concatenate(results_boxes), columns=['left', 'top', 'width', 'height'])
score_df = pd.DataFrame({'scores':np.concatenate(results_scores),'labels':np.concatenate(results_labels), 'image_name':result_image_ids})
test_df = pd.concat([box_df, score_df], axis=1 ) | NFL 1st and Future - Impact Detection |
14,018,577 | for dirname, _, filenames in os.walk('/kaggle/input'):
for filename in filenames:
print(os.path.join(dirname, filename))
train = pd.read_csv('.. /input/titanic/train.csv')
test = pd.read_csv('.. /input/titanic/test.csv')
print('Shape of train dataset: {}'.format(train.shape))
print('Shape of test dataset: {}'.format(test.shape))<concatenate> | test_df['gameKey'] = test_df.image_name.str.split('_' ).str[0].astype(int)
test_df['playID'] = test_df.image_name.str.split('_' ).str[1].astype(int)
test_df['view'] = test_df.image_name.str.split('_' ).str[2]
test_df['frame'] = test_df.image_name.str.split('_' ).str[3].str.replace('.png','' ).astype(int)
test_df['video'] = test_df.image_name.str.rsplit('_',1 ).str[0] + '.mp4'
| NFL 1st and Future - Impact Detection |
14,018,577 | train['Type'] = 'train'
test['Type'] = 'test'
all = pd.concat([train, test], sort=False ).reset_index(drop=True)
print('Shape of all dataset: {}'.format(all.shape))<count_missing_values> | test_df.to_csv('off.csv',index=False ) | NFL 1st and Future - Impact Detection |
14,018,577 | print(all.isnull().values.any())
train.isnull().sum()<filter> | import numpy as np
import pandas as pd
from scipy.optimize import linear_sum_assignment | NFL 1st and Future - Impact Detection |
14,018,577 | all_corr[all_corr['level_0'] == 'Age']<groupby> | SEED = 42
FOLDS = 5
FOLD = 2
EFF_DET = 5
Threshold = 0.45
LOOK_BACKWARD_LIMIT = 10
IOU_THRESHOLD = 0.4
IMAGE_SIZE = 512
NUM_CLASSES = 2
BATCH_SIZE = 16
FRAME_THRESHOLD = 120
keep_columns_threshold = 20 | NFL 1st and Future - Impact Detection |
14,018,577 | all['Age'] = all.groupby(['Pclass', 'SibSp'])['Age'].apply(lambda x: x.fillna(x.median()))<feature_engineering> | def iou(bbox1, bbox2):
bbox1 = [float(x)for x in bbox1]
bbox2 = [float(x)for x in bbox2]
(x0_1, y0_1, x1_1, y1_1)= bbox1
(x0_2, y0_2, x1_2, y1_2)= bbox2
overlap_x0 = max(x0_1, x0_2)
overlap_y0 = max(y0_1, y0_2)
overlap_x1 = min(x1_1, x1_2)
overlap_y1 = min(y1_1, y1_2)
if overlap_x1 - overlap_x0 <= 0 or overlap_y1 - overlap_y0 <= 0:
return 0
size_1 =(x1_1 - x0_1)*(y1_1 - y0_1)
size_2 =(x1_2 - x0_2)*(y1_2 - y0_2)
size_intersection =(overlap_x1 - overlap_x0)*(overlap_y1 - overlap_y0)
size_union = size_1 + size_2 - size_intersection
return size_intersection / size_union
def precision_calc(gt_boxes, pred_boxes):
cost_matix = np.ones(( len(gt_boxes), len(pred_boxes)))
for i, box1 in enumerate(gt_boxes):
for j, box2 in enumerate(pred_boxes):
dist = abs(box1[0]-box2[0])
if dist > 4:
continue
iou_score = iou(box1[1:], box2[1:])
if iou_score < 0.35:
continue
else:
cost_matix[i,j]=0
row_ind, col_ind = linear_sum_assignment(cost_matix)
fn = len(gt_boxes)- row_ind.shape[0]
fp = len(pred_boxes)- col_ind.shape[0]
tp=0
for i, j in zip(row_ind, col_ind):
if cost_matix[i,j]==0:
tp+=1
else:
fp+=1
fn+=1
return tp, fp, fn | NFL 1st and Future - Impact Detection |
14,018,577 | print(all_corr[all_corr['level_0'] == 'Fare'])
all['Fare'] = all.groupby(['Pclass', 'Parch', 'SibSp'])['Fare'].apply(lambda x: x.fillna(x.median()))<count_values> | def filter_duplicates_and_threshols(row):
boxes,scores,labels = row.box_predictions,row.score_list,row.label_list
filtered_boxes = []
filtered_scores = []
filtered_labels = []
for box,score,label in zip(boxes,scores,labels):
if score> Threshold:
add_box = True
for i,ref_box in enumerate(filtered_boxes):
if iou(box,ref_box)>0.5:
add_box = False
if label==2:filtered_labels[i]=2
if add_box:
filtered_boxes.append(box)
filtered_scores.append(score)
filtered_labels.append(label)
row.box_predictions,row.score_list,row.label_list = filtered_boxes,filtered_scores,filtered_labels
return row
def filter_duplicates_and_low_scores(row):
boxes,scores,labels = row.box_predictions,row.score_list,row.label_list
ordered_sample_predictions = sorted(list(zip(boxes,scores,labels)) , key = lambda x:x[1],
reverse = True)
filtered_boxes = []
filtered_scores = []
filtered_labels = []
for box,score,label in ordered_sample_predictions:
if score>Threshold:
add_box = True
for i,ref_box in enumerate(filtered_boxes):
if iou(box,ref_box)>0.5:
add_box = False
if label==2:filtered_labels[i]=2
if add_box:
filtered_boxes.append(box)
filtered_scores.append(score)
filtered_labels.append(label)
row.box_predictions,row.score_list,row.label_list = filtered_boxes,filtered_scores,filtered_labels
return row | NFL 1st and Future - Impact Detection |
14,018,577 | all['Cabin'].value_counts()<data_type_conversions> | video_labels = pd.read_csv('off.csv' ) | NFL 1st and Future - Impact Detection |
14,018,577 | all['Cabin'] = all['Cabin'].fillna('N' )<count_values> | video_labels = video_labels.sort_values(by=['video','frame'] ).reset_index(drop=True)
video_labels = video_labels[video_labels.frame>0]
video_labels['image'] = video_labels['video'].str.replace('.mp4', '')+ '_' + video_labels['frame'].astype(str)+ '.png'
video_labels['right'] = video_labels['left'] + video_labels['width']
video_labels['bottom'] = video_labels['top'] + video_labels['height']
video_labels['label_list'] = video_labels['labels'].apply(lambda x:[x])
video_labels['bbox'] = pd.Series(video_labels[['left','top','right','bottom']].values.tolist() , index=video_labels.index)
video_labels['box_predictions'] = video_labels['bbox'].apply(lambda x:[x])
video_labels['score_list'] = video_labels['scores'].apply(lambda x:[x])
base_cols = ['video','image','gameKey','frame']
target_cols = ['box_predictions','label_list','score_list']
predictions = video_labels[base_cols+target_cols].groupby(base_cols ).sum().reset_index()
predictions = predictions.apply(filter_duplicates_and_low_scores,axis=1)
predictions['n_impact'] = predictions.label_list.apply(lambda x:(np.array(x)==2 ).sum())
predictions.groupby('video' ).n_impact.sum().describe() | NFL 1st and Future - Impact Detection |
14,018,577 | print(all['Embarked'].value_counts())
all['Embarked'] = all['Embarked'].fillna('S' )<count_missing_values> | def iou(bbox1, bbox2):
bbox1 = [float(x)for x in bbox1]
bbox2 = [float(x)for x in bbox2]
(x0_1, y0_1, x1_1, y1_1)= bbox1
(x0_2, y0_2, x1_2, y1_2)= bbox2
overlap_x0 = max(x0_1, x0_2)
overlap_y0 = max(y0_1, y0_2)
overlap_x1 = min(x1_1, x1_2)
overlap_y1 = min(y1_1, y1_2)
if overlap_x1 - overlap_x0 <= 0 or overlap_y1 - overlap_y0 <= 0:
return 0
size_1 =(x1_1 - x0_1)*(y1_1 - y0_1)
size_2 =(x1_2 - x0_2)*(y1_2 - y0_2)
size_intersection =(overlap_x1 - overlap_x0)*(overlap_y1 - overlap_y0)
size_union = size_1 + size_2 - size_intersection
return size_intersection / size_union
def match_boxes(bbox_set1,bbox_set2,diagonal=True,iou_threshold=0.35):
all_pairs = {}
relation = [None for x in bbox_set2]
done1 = {}
done2 = {}
for i,bbox1 in enumerate(bbox_set1):
for j,bbox2 in enumerate(bbox_set2):
if i!=j or diagonal:
all_pairs[(i,j)] = iou(bbox1,bbox2)
for(index1,index2),iou_score in sorted(all_pairs.items() , key=lambda item: item[1],reverse=True):
if iou_score<iou_threshold: return relation
if(index1 not in done1)and(index2 not in done2):
relation[index2] = index1
done1[index1] = 1
done2[index2] = 1
return relation
def map_classes(row):
mapping = {}
for box,pred in zip(row.box_predictions,row.label_list):
mapping[tuple(box)] = pred
return mapping
def assign_players(df,video,LOOK_BACKWARD_LIMIT=LOOK_BACKWARD_LIMIT,IOU_THRESHOLD=IOU_THRESHOLD):
video_df = df[df.video==video].sort_values(by='frame' ).set_index('frame')
row_to_class_map = video_df['row_to_class_map']
row_to_class_map.apply(lambda x:x.update({() :-1}))
player_to_bbox = pd.DataFrame()
bbox_to_player = {k+1:{} for k in range(video_df.shape[0])}
N_PLAYERS = 0
bboxes = video_df.loc[1,'box_predictions']
for player_id,box in enumerate(bboxes):
player_to_bbox[player_id] = None
player_to_bbox.at[1,player_id] = box
bbox_to_player[1][tuple(box)] = player_id
N_PLAYERS += 1
for frame in range(2,video_df.shape[0]+1):
player_to_bbox.loc[frame] = None
bboxes = video_df.loc[frame,'box_predictions']
assigned = {}
look_backward = 1
while(look_backward<min(frame,LOOK_BACKWARD_LIMIT+1)and len(bboxes)>0):
ref_bbox = []
for box in video_df.loc[frame-look_backward,'box_predictions']:
if bbox_to_player[frame-look_backward][tuple(box)] not in assigned:
ref_bbox.append(box)
relation = match_boxes(ref_bbox,bboxes,iou_threshold=IOU_THRESHOLD)
left_out_bboxes = []
for index,ref_index in enumerate(relation):
if ref_index is not None:
player_id = bbox_to_player[frame-look_backward][tuple(ref_bbox[ref_index])]
box = bboxes[index]
player_to_bbox.at[frame,player_id] = box
bbox_to_player[frame][tuple(box)] = player_id
assigned[player_id] = True
else:
left_out_bboxes.append(bboxes[index])
bboxes = left_out_bboxes
look_backward +=1
for box in bboxes:
player_id = N_PLAYERS
player_to_bbox[player_id] = None
player_to_bbox.at[frame,player_id] = box
bbox_to_player[frame][tuple(box)] = player_id
N_PLAYERS += 1
player_to_class_map = player_to_bbox.apply(lambda row: row.fillna('' ).apply(tuple ).apply(
lambda x:row_to_class_map.loc[row.name][x]),axis=1)
return player_to_bbox,player_to_class_map
def filter_impacts(player_series):
impacted_player_series = pd.DataFrame(player_series.loc[player_series==2])
impacted_player_series['indices'] = impacted_player_series.index
start_indices = impacted_player_series[impacted_player_series.indices.diff(1 ).fillna(99 ).abs() >9].indices.tolist()
end_indices = impacted_player_series[impacted_player_series.indices.diff(-1 ).fillna(99 ).abs() >9].indices.tolist()
output_series = player_series.copy(deep=True)
output_series[player_series>0] = 1
output_series[player_series<0] = -1
for start_index,end_index in zip(start_indices,end_indices):
difference = end_index-start_index+1
n_medians = math.ceil(difference/9)
difference_split = difference/(n_medians+1)
centres = [int(start_index+(i+1)*difference_split)for i in range(n_medians)]
output_series.loc[centres] = 2
return output_series | NFL 1st and Future - Impact Detection |
14,018,577 | all.isnull().values.any()<string_transform> | submission_df = pd.DataFrame(columns=['gameKey','playID','view','video','frame','left','width','top','height'])
predictions['row_to_class_map'] = predictions.apply(map_classes,axis=1)
entries = 0
for video in predictions.video.unique() :
assigned_players_boxes,assigned_players_classes = assign_players(predictions,video)
keep_columns = assigned_players_classes.columns[(assigned_players_classes!=-1 ).sum() >keep_columns_threshold]
assigned_players_boxes = assigned_players_boxes[keep_columns]
assigned_players_classes = assigned_players_classes[keep_columns]
print(assigned_players_boxes.shape)
assigned_players_classes_filtered = assigned_players_classes.apply(filter_impacts)
n_filtered =(assigned_players_classes_filtered==2 ).sum().sum()
n_classes =(assigned_players_classes==2 ).sum().sum()
print(video,n_filtered,n_classes,(~assigned_players_boxes.isna() ).sum().sum())
all_bboxes = assigned_players_boxes.fillna(method='ffill' ).values.reshape(-1)
all_classes = assigned_players_classes_filtered.values.reshape(-1)
all_frames = assigned_players_boxes.apply(lambda x: pd.DataFrame(x ).apply(
lambda y: y.name, axis=1)).values.reshape(-1)
bboxes_with_impact = all_bboxes[np.where(all_classes==2)]
frames_with_impact = all_frames[np.where(all_classes==2)]
gameKey,playID,view,_ = video.replace('.','_' ).split('_')
for frame,box in zip(frames_with_impact,bboxes_with_impact):
submission_df.loc[entries,'gameKey'] = gameKey
submission_df.loc[entries,'playID'] = playID
submission_df.loc[entries,'view'] = view
submission_df.loc[entries,'video'] = video
submission_df.loc[entries,'frame'] = frame
submission_df.loc[entries,'left'] = box[0]
submission_df.loc[entries,'top'] = box[1]
submission_df.loc[entries,'width'] = box[2]-box[0]
submission_df.loc[entries,'height'] = box[3]-box[1]
entries += 1
submission_df = submission_df[(submission_df.frame<FRAME_THRESHOLD)&(submission_df.frame>20)] | NFL 1st and Future - Impact Detection |
14,018,577 | def extract_surname(data):
families = []
for i in range(len(data)) :
name = data.iloc[i]
if '(' in name:
name_no_bracket = name.split('(')[0]
else:
name_no_bracket = name
family = name_no_bracket.split(',')[0]
title = name_no_bracket.split(',')[1].strip().split(' ')[0]
for c in string.punctuation:
family = family.replace(c, '' ).strip()
families.append(family)
return families
print(all['Name'].value_counts())
all['Title'] = all['Name'].str.split(', ', expand=True)[1].str.split('.', expand=True)[0]
all['Family'] = extract_surname(all['Name'])
all['IsMarried'] = np.where(all['Title'] == 'Mrs', 1, 0)
all.drop(['Name'], inplace=True, axis=1 )<feature_engineering> | submission_df1 = submission_df.copy(deep=True ) | NFL 1st and Future - Impact Detection |
14,018,577 | all['FamilySize'] = all['SibSp'] + all['Parch'] + 1<categorify> | SEED = 42
FOLDS = 5
FOLD = 2
EFF_DET = 5
Threshold = 0.65
LOOK_BACKWARD_LIMIT = 10
IOU_THRESHOLD = 0.4
IMAGE_SIZE = 512
NUM_CLASSES = 2
BATCH_SIZE = 16
FRAME_THRESHOLD = 240
keep_columns_threshold = 20
video_labels = pd.read_csv('off.csv')
video_labels = video_labels.sort_values(by=['video','frame'] ).reset_index(drop=True)
video_labels = video_labels[video_labels.frame>0]
video_labels['image'] = video_labels['video'].str.replace('.mp4', '')+ '_' + video_labels['frame'].astype(str)+ '.png'
video_labels['right'] = video_labels['left'] + video_labels['width']
video_labels['bottom'] = video_labels['top'] + video_labels['height']
video_labels['label_list'] = video_labels['labels'].apply(lambda x:[x])
video_labels['bbox'] = pd.Series(video_labels[['left','top','right','bottom']].values.tolist() , index=video_labels.index)
video_labels['box_predictions'] = video_labels['bbox'].apply(lambda x:[x])
video_labels['score_list'] = video_labels['scores'].apply(lambda x:[x])
base_cols = ['video','image','gameKey','frame']
target_cols = ['box_predictions','label_list','score_list']
predictions = video_labels[base_cols+target_cols].groupby(base_cols ).sum().reset_index()
predictions = predictions.apply(filter_duplicates_and_low_scores,axis=1)
predictions['n_impact'] = predictions.label_list.apply(lambda x:(np.array(x)==2 ).sum())
predictions.groupby('video' ).n_impact.sum().describe()
submission_df = pd.DataFrame(columns=['gameKey','playID','view','video','frame','left','width','top','height'])
predictions['row_to_class_map'] = predictions.apply(map_classes,axis=1)
entries = 0
for video in predictions.video.unique() :
assigned_players_boxes,assigned_players_classes = assign_players(predictions,video)
keep_columns = assigned_players_classes.columns[(assigned_players_classes!=-1 ).sum() >keep_columns_threshold]
assigned_players_boxes = assigned_players_boxes[keep_columns]
assigned_players_classes = assigned_players_classes[keep_columns]
print(assigned_players_boxes.shape)
assigned_players_classes_filtered = assigned_players_classes.apply(filter_impacts)
n_filtered =(assigned_players_classes_filtered==2 ).sum().sum()
n_classes =(assigned_players_classes==2 ).sum().sum()
print(video,n_filtered,n_classes,(~assigned_players_boxes.isna() ).sum().sum())
all_bboxes = assigned_players_boxes.fillna(method='ffill' ).values.reshape(-1)
all_classes = assigned_players_classes_filtered.values.reshape(-1)
all_frames = assigned_players_boxes.apply(lambda x: pd.DataFrame(x ).apply(
lambda y: y.name, axis=1)).values.reshape(-1)
bboxes_with_impact = all_bboxes[np.where(all_classes==2)]
frames_with_impact = all_frames[np.where(all_classes==2)]
gameKey,playID,view,_ = video.replace('.','_' ).split('_')
for frame,box in zip(frames_with_impact,bboxes_with_impact):
submission_df.loc[entries,'gameKey'] = gameKey
submission_df.loc[entries,'playID'] = playID
submission_df.loc[entries,'view'] = view
submission_df.loc[entries,'video'] = video
submission_df.loc[entries,'frame'] = frame
submission_df.loc[entries,'left'] = box[0]
submission_df.loc[entries,'top'] = box[1]
submission_df.loc[entries,'width'] = box[2]-box[0]
submission_df.loc[entries,'height'] = box[3]-box[1]
entries += 1
submission_df = submission_df[(submission_df.frame<FRAME_THRESHOLD)&(submission_df.frame>120)] | NFL 1st and Future - Impact Detection |
14,018,577 | family_map = {1: 'Alone', 2: 'Small', 3: 'Small', 4: 'Small', 5: 'Medium', 6: 'Medium', 7: 'Large', 8: 'Large', 11: 'Large'}
all['FamilySizeGrouped'] = all['FamilySize'].map(family_map )<feature_engineering> | submission_df2 = submission_df.copy(deep=True ) | NFL 1st and Future - Impact Detection |
14,018,577 | all['Ticket_count'] = all.Ticket.apply(lambda x: all[all['Ticket']==x].shape[0] )<feature_engineering> | submission_df = pd.concat([submission_df1,submission_df2] ) | NFL 1st and Future - Impact Detection |
14,018,577 | all.loc[(all['Sex'] == 'male')&(all['Age'] > 18.0), 'Sex'] = 0
all.loc[(all['Sex'] == 'male')&(all['Age'] <= 18.0), 'Sex'] = 1
all.loc[all['Sex'] == 'female', 'Sex'] = 2<feature_engineering> | env = nflimpact.make_env() | NFL 1st and Future - Impact Detection |
14,018,577 | <count_values><EOS> | if IS_PRIVATE:
env.predict(submission_df)
else:
sub = pd.read_csv('.. /input/nfl-impact-detection/sample_submission.csv')
env.predict(sub ) | NFL 1st and Future - Impact Detection |
10,378,659 | <SOS> metric: macrofscore Kaggle data source: petals-to-the-metal-flower-classification-on-tpu<predict_on_test> | !pip install -q efficientnet | Petals to the Metal - Flower Classification on TPU |
10,378,659 | stack_ypred = stacking_clf.predict(dval_X)
print('Stacking classifier accuracy score is.... ',accuracy_score(val_Y,stack_ypred))<predict_on_test> | print('TensorFlow version', tf.__version__)
AUTO = tf.data.experimental.AUTOTUNE | Petals to the Metal - Flower Classification on TPU |
10,378,659 | prediction_stacking_clf = stacking_clf.predict(X_test )<save_to_csv> | start_time = datetime.now()
print('Time now is', start_time)
end_training_by_tdelta = timedelta(seconds=8400)
this_run_file_prefix = start_time.strftime('%Y%m%d_%H%M_')
print(this_run_file_prefix)
IMAGE_SIZE = [224, 224]
EPOCHS = 12
BATCH_SIZE = 16 * strategy.num_replicas_in_sync
GCS_PATH_SELECT = {
192: GCS_DS_PATH + '/tfrecords-jpeg-192x192',
224: GCS_DS_PATH + '/tfrecords-jpeg-224x224',
331: GCS_DS_PATH + '/tfrecords-jpeg-331x331',
512: GCS_DS_PATH + '/tfrecords-jpeg-512x512'
}
GCS_PATH = GCS_PATH_SELECT[IMAGE_SIZE[0]]
TRAINING_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/train/*.tfrec')
VALIDATION_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/val/*.tfrec')
TEST_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/test/*.tfrec')
MOREIMAGES_PATH_SELECT = {
192: '/tfrecords-jpeg-192x192',
224: '/tfrecords-jpeg-224x224',
331: '/tfrecords-jpeg-331x331',
512: '/tfrecords-jpeg-512x512'
}
MOREIMAGES_PATH = MOREIMAGES_PATH_SELECT[IMAGE_SIZE[0]]
IMAGENET_FILES = tf.io.gfile.glob(MORE_IMAGES_GCS_DS_PATH + '/imagenet' + MOREIMAGES_PATH + '/*.tfrec')
INATURELIST_FILES = tf.io.gfile.glob(MORE_IMAGES_GCS_DS_PATH + '/inaturalist' + MOREIMAGES_PATH + '/*.tfrec')
OPENIMAGE_FILES = tf.io.gfile.glob(MORE_IMAGES_GCS_DS_PATH + '/openimage' + MOREIMAGES_PATH + '/*.tfrec')
OXFORD_FILES = tf.io.gfile.glob(MORE_IMAGES_GCS_DS_PATH + '/oxford_102' + MOREIMAGES_PATH + '/*.tfrec')
TENSORFLOW_FILES = tf.io.gfile.glob(MORE_IMAGES_GCS_DS_PATH + '/tf_flowers' + MOREIMAGES_PATH + '/*.tfrec')
ADDITIONAL_TRAINING_FILENAMES = IMAGENET_FILES + INATURELIST_FILES + OPENIMAGE_FILES + OXFORD_FILES + TENSORFLOW_FILES
print('----')
TRAINING_FILENAMES = TRAINING_FILENAMES + ADDITIONAL_TRAINING_FILENAMES + VALIDATION_FILENAMES
CLASSES = ['pink primrose', 'hard-leaved pocket orchid', 'canterbury bells', 'sweet pea', 'wild geranium', 'tiger lily', 'moon orchid', 'bird of paradise', 'monkshood', 'globe thistle',
'snapdragon', "colt's foot", 'king protea', 'spear thistle', 'yellow iris', 'globe-flower', 'purple coneflower', 'peruvian lily', 'balloon flower', 'giant white arum lily',
'fire lily', 'pincushion flower', 'fritillary', 'red ginger', 'grape hyacinth', 'corn poppy', 'prince of wales feathers', 'stemless gentian', 'artichoke', 'sweet william',
'carnation', 'garden phlox', 'love in the mist', 'cosmos', 'alpine sea holly', 'ruby-lipped cattleya', 'cape flower', 'great masterwort', 'siam tulip', 'lenten rose',
'barberton daisy', 'daffodil', 'sword lily', 'poinsettia', 'bolero deep blue', 'wallflower', 'marigold', 'buttercup', 'daisy', 'common dandelion',
'petunia', 'wild pansy', 'primula', 'sunflower', 'lilac hibiscus', 'bishop of llandaff', 'gaura', 'geranium', 'orange dahlia', 'pink-yellow dahlia',
'cautleya spicata', 'japanese anemone', 'black-eyed susan', 'silverbush', 'californian poppy', 'osteospermum', 'spring crocus', 'iris', 'windflower', 'tree poppy',
'gazania', 'azalea', 'water lily', 'rose', 'thorn apple', 'morning glory', 'passion flower', 'lotus', 'toad lily', 'anthurium',
'frangipani', 'clematis', 'hibiscus', 'columbine', 'desert-rose', 'tree mallow', 'magnolia', 'cyclamen ', 'watercress', 'canna lily',
'hippeastrum ', 'bee balm', 'pink quill', 'foxglove', 'bougainvillea', 'camellia', 'mallow', 'mexican petunia', 'bromelia', 'blanket flower',
'trumpet creeper', 'blackberry lily', 'common tulip', 'wild rose'] | Petals to the Metal - Flower Classification on TPU |
10,378,659 | output = pd.DataFrame({"PassengerId":passengerid , "Survived" : prediction_stacking_clf})
output.to_csv("Submission_stacking_clf.csv",index = False )<load_from_csv> | LR_START = 0.00001
LR_MAX = 0.00005 * strategy.num_replicas_in_sync
LR_MIN = LR_START
LR_RAMPUP_EPOCHS = 5
LR_SUSTAIN_EPOCHS = 0
LR_EXP_DECAY = 0.80
def lrfn(epoch):
if epoch < LR_RAMPUP_EPOCHS:
lr = LR_START +(epoch *(LR_MAX - LR_START)/ LR_RAMPUP_EPOCHS)
elif epoch <(LR_RAMPUP_EPOCHS + LR_SUSTAIN_EPOCHS):
lr = LR_MAX
else:
lr = LR_MIN +(LR_MAX - LR_MIN)* LR_EXP_DECAY **(epoch - LR_RAMPUP_EPOCHS - LR_SUSTAIN_EPOCHS)
return lr
lr_callback = tf.keras.callbacks.LearningRateScheduler(lrfn, verbose = True)
rng = [i for i in range(30)]
y = [lrfn(x)for x in rng]
plt.plot(rng, y)
print(y ) | Petals to the Metal - Flower Classification on TPU |
10,378,659 | df_train = pd.read_csv('.. /input/train.csv')
df_test = pd.read_csv('.. /input/test.csv')
df_train.head()<prepare_output> | np.set_printoptions(threshold=15, linewidth=80)
def batch_to_numpy_images_and_labels(data):
images, labels = data
numpy_images = images.numpy()
numpy_labels = labels.numpy()
if numpy_labels.dtype == object:
numpy_labels = [None for _ in enumerate(numpy_images)]
return numpy_images, numpy_labels
def title_from_label_and_target(label, correct_label):
if correct_label is None:
return CLASSES[label], True
correct =(label == correct_label)
return "{} [{}{}{}]".format(CLASSES[label], 'OK' if correct else 'NO', u"\u2192" if not correct else '',
CLASSES[correct_label] if not correct else ''), correct
def display_one_flower(image, title, subplot, red=False, titlesize=16):
plt.subplot(*subplot)
plt.axis('off')
plt.imshow(image)
if len(title)> 0:
plt.title(title, fontsize=int(titlesize)if not red else int(titlesize/1.2), color='red' if red else 'black', fontdict={'verticalalignment':'center'}, pad=int(titlesize/1.5))
return(subplot[0], subplot[1], subplot[2]+1)
def display_batch_of_images(databatch, predictions=None):
images, labels = batch_to_numpy_images_and_labels(databatch)
if labels is None:
labels = [None for _ in enumerate(images)]
rows = int(math.sqrt(len(images)))
cols = len(images)//rows
FIGSIZE = 13.0
SPACING = 0.1
subplot=(rows,cols,1)
if rows < cols:
plt.figure(figsize=(FIGSIZE,FIGSIZE/cols*rows))
else:
plt.figure(figsize=(FIGSIZE/rows*cols,FIGSIZE))
for i,(image, label)in enumerate(zip(images[:rows*cols], labels[:rows*cols])) :
title = '' if label is None else CLASSES[label]
correct = True
if predictions is not None:
title, correct = title_from_label_and_target(predictions[i], label)
dynamic_titlesize = FIGSIZE*SPACING/max(rows,cols)*40+3
subplot = display_one_flower(image, title, subplot, not correct, titlesize=dynamic_titlesize)
plt.tight_layout()
if label is None and predictions is None:
plt.subplots_adjust(wspace=0, hspace=0)
else:
plt.subplots_adjust(wspace=SPACING, hspace=SPACING)
plt.show()
def display_confusion_matrix(cmat, score, precision, recall):
plt.figure(figsize=(15,15))
ax = plt.gca()
ax.matshow(cmat, cmap='Reds')
ax.set_xticks(range(len(CLASSES)))
ax.set_xticklabels(CLASSES, fontdict={'fontsize': 7})
plt.setp(ax.get_xticklabels() , rotation=45, ha="left", rotation_mode="anchor")
ax.set_yticks(range(len(CLASSES)))
ax.set_yticklabels(CLASSES, fontdict={'fontsize': 7})
plt.setp(ax.get_yticklabels() , rotation=45, ha="right", rotation_mode="anchor")
titlestring = ""
if score is not None:
titlestring += 'f1 = {:.3f} '.format(score)
if precision is not None:
titlestring += '
precision = {:.3f} '.format(precision)
if recall is not None:
titlestring += '
recall = {:.3f} '.format(recall)
if len(titlestring)> 0:
ax.text(101, 1, titlestring, fontdict={'fontsize': 18, 'horizontalalignment':'right', 'verticalalignment':'top', 'color':'
plt.show()
def display_training_curves(training, validation, title, subplot):
if subplot%10==1:
plt.subplots(figsize=(10,10), facecolor='
plt.tight_layout()
ax = plt.subplot(subplot)
ax.set_facecolor('
ax.plot(training)
ax.plot(validation)
ax.set_title('model '+ title)
ax.set_ylabel(title)
ax.set_xlabel('epoch')
ax.legend(['train', 'valid.'] ) | Petals to the Metal - Flower Classification on TPU |
10,378,659 | df_train.set_index('Id', inplace=True)
df_test.set_index('Id', inplace=True )<drop_column> | def decode_image(image_data):
image = tf.image.decode_jpeg(image_data, channels = 3)
image = tf.cast(image, tf.float32)/ 255.0
image = tf.reshape(image, [*IMAGE_SIZE, 3])
return image
def read_labeled_tfrecord(example):
LABELED_TFREC_FORMAT = {
'image': tf.io.FixedLenFeature([], tf.string),
'class': tf.io.FixedLenFeature([], tf.int64),
}
example = tf.io.parse_single_example(example, LABELED_TFREC_FORMAT)
image = decode_image(example['image'])
label = tf.cast(example['class'], tf.int32)
return image, label
def read_unlabeled_tfrecord(example):
UNLABELED_TFREC_FORMAT = {
'image': tf.io.FixedLenFeature([], tf.string),
'id': tf.io.FixedLenFeature([], tf.string),
}
example = tf.io.parse_single_example(example, UNLABELED_TFREC_FORMAT)
image = decode_image(example['image'])
idnum = example['id']
return image, idnum
def load_dataset(filenames, labeled = True, ordered = False):
ignore_order = tf.data.Options()
if not ordered:
ignore_order.experimental_deterministic = False
dataset = tf.data.TFRecordDataset(filenames, num_parallel_reads = AUTO)
dataset = dataset.with_options(ignore_order)
dataset = dataset.map(read_labeled_tfrecord if labeled else read_unlabeled_tfrecord, num_parallel_calls = AUTO)
return dataset
def data_augment(image, label):
image = tf.image.random_flip_left_right(image)
image = tf.image.random_saturation(image, 0, 2)
return image, label
def get_training_dataset() :
dataset = load_dataset(TRAINING_FILENAMES, labeled = True)
dataset = dataset.map(data_augment, num_parallel_calls = AUTO)
dataset = dataset.repeat()
dataset = dataset.shuffle(2048)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.prefetch(AUTO)
return dataset
def get_validation_dataset(ordered = False):
dataset = load_dataset(VALIDATION_FILENAMES, labeled = True, ordered=ordered)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.cache()
dataset = dataset.prefetch(AUTO)
return dataset
def get_test_dataset(ordered = False):
dataset = load_dataset(TEST_FILENAMES, labeled = False, ordered = ordered)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.prefetch(AUTO)
return dataset
def count_data_items(filenames):
n = [int(re.compile(r"-([0-9]*)\." ).search(filename ).group(1)) for filename in filenames]
return np.sum(n)
NUM_TRAINING_IMAGES = count_data_items(TRAINING_FILENAMES)
NUM_VALIDATION_IMAGES = count_data_items(VALIDATION_FILENAMES)
NUM_TEST_IMAGES = count_data_items(TEST_FILENAMES)
STEPS_PER_EPOCH = NUM_TRAINING_IMAGES // BATCH_SIZE
print('Dataset: {} training images, {} validation images, {} unlabeled test images'.format(NUM_TRAINING_IMAGES, NUM_VALIDATION_IMAGES, NUM_TEST_IMAGES)) | Petals to the Metal - Flower Classification on TPU |
10,378,659 | df_train.drop(['GarageArea','1stFlrSF','TotRmsAbvGrd','2ndFlrSF'], axis=1, inplace=True)
df_test.drop(['GarageArea','1stFlrSF','TotRmsAbvGrd','2ndFlrSF'], axis=1, inplace=True )<filter> | training_dataset = get_training_dataset()
training_dataset = training_dataset.unbatch().batch(20)
train_batch = iter(training_dataset)
| Petals to the Metal - Flower Classification on TPU |
10,378,659 | df_train = df_train[df_train['GrLivArea']<4500]<count_missing_values> | test_dataset = get_test_dataset()
test_dataset = test_dataset.unbatch().batch(20)
test_batch = iter(test_dataset)
| Petals to the Metal - Flower Classification on TPU |
10,378,659 | def check_nulls(df):
percent_missing =(df.isnull().sum() * 100 / len(df)).sort_values()
return round(percent_missing,2 )<count_missing_values> | test_dataset = get_test_dataset()
test_dataset = test_dataset.unbatch().batch(20)
test_batch = iter(test_dataset ) | Petals to the Metal - Flower Classification on TPU |
10,378,659 | check_nulls(df_train )<count_missing_values> | early_stop = tf.keras.callbacks.EarlyStopping(monitor='val_loss', patience=20, restore_best_weights = True ) | Petals to the Metal - Flower Classification on TPU |
10,378,659 | check_nulls(df_test )<define_variables> | def create_VGG16_model() :
pretrained_model = tf.keras.applications.VGG16(weights = 'imagenet', include_top = False, input_shape = [*IMAGE_SIZE, 3])
pretrained_model.trainable = True
model = tf.keras.Sequential([
pretrained_model,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation = 'softmax')
])
model.compile(optimizer = 'adam', loss = 'sparse_categorical_crossentropy', metrics = ['sparse_categorical_accuracy'])
return model | Petals to the Metal - Flower Classification on TPU |
10,378,659 | categorical_list = [col for col in df_train.columns if df_train[col].dtypes == object]
numerical_list = [col for col in df_train.columns if df_train[col].dtypes != object]
print('Categories:', categorical_list)
print('Numbers:', numerical_list )<categorify> | def create_Xception_model() :
pretrained_model = tf.keras.applications.Xception(include_top = False, input_shape = [*IMAGE_SIZE, 3])
pretrained_model.trainable = True
model = tf.keras.Sequential([
pretrained_model,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation = 'softmax')
])
model.compile(optimizer = 'adam', loss = 'sparse_categorical_crossentropy', metrics = ['sparse_categorical_accuracy'])
return model | Petals to the Metal - Flower Classification on TPU |
10,378,659 | def fill_missing_values(df):
lst = ["Alley","BsmtQual","BsmtCond","BsmtExposure","BsmtFinType1",
"BsmtFinType2","Fence","FireplaceQu","GarageType","GarageFinish",
"GarageQual","GarageCond","Electrical","GarageFinish","MiscFeature","MasVnrType","PoolQC"]
for col in lst:
df[col] = df[col].fillna("Not present")
lst = ['GarageYrBlt','MasVnrArea','BsmtFinSF1','BsmtFinSF2','TotalBsmtSF',
'BsmtUnfSF','BsmtFullBath','BsmtHalfBath','MasVnrArea','GarageCars']
for col in lst:
df[col] = df[col].fillna(0)
lst = ['Utilities','MSZoning','Exterior1st','Exterior2nd','Electrical','KitchenQual']
for col in lst:
df[col] = df[col].fillna(df[col].mode() [0])
df['Functional'] = df['Functional'].fillna('Typ')
df['SaleType'] = df['SaleType'].fillna('Normal')
df['LotFrontage'] = df['LotFrontage'].fillna(df.LotFrontage.mean())
df.drop('PoolQC', axis=1, inplace=True)
<count_missing_values> | def create_DenseNet_model() :
pretrained_model = tf.keras.applications.DenseNet201(weights = 'imagenet', include_top = False, input_shape = [*IMAGE_SIZE, 3])
pretrained_model.trainable = True
model = tf.keras.Sequential([
pretrained_model,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation = 'softmax')
])
model.compile(optimizer = 'adam', loss = 'sparse_categorical_crossentropy', metrics = ['sparse_categorical_accuracy'])
return model | Petals to the Metal - Flower Classification on TPU |
10,378,659 | fill_missing_values(df_train)
fill_missing_values(df_test )<count_missing_values> | def create_EfficientNet_model() :
pretrained_model = efficientnet.EfficientNetB7(weights = 'noisy-student', include_top = False, input_shape = [*IMAGE_SIZE, 3])
pretrained_model.trainable = True
model = tf.keras.Sequential([
pretrained_model,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation = 'softmax')
])
model.compile(optimizer = 'adam', loss = 'sparse_categorical_crossentropy', metrics = ['sparse_categorical_accuracy'])
return model | Petals to the Metal - Flower Classification on TPU |
10,378,659 | print(df_train.isnull().sum().sum())
print(df_test.isnull().sum().sum() )<categorify> | def create_InceptionV3_model() :
pretrained_model = tf.keras.applications.InceptionV3(weights = 'imagenet', include_top = False, input_shape = [*IMAGE_SIZE, 3])
pretrained_model.trainable = True
model = tf.keras.Sequential([
pretrained_model,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation = 'softmax')
])
model.compile(optimizer = 'adam', loss = 'sparse_categorical_crossentropy', metrics = ['sparse_categorical_accuracy'])
return model | Petals to the Metal - Flower Classification on TPU |
10,378,659 | def label_encode(df):
df['MSSubClass'] = df['MSSubClass'].astype(object)
df = df.replace({"Alley" : {"Not present" : 0, "Grvl" : 1, "Pave" : 2},
"BsmtCond" : {"Not present" : 0, "Po" : 1, "Fa" : 2, "TA" : 3, "Gd" : 4, "Ex" : 5},
"BsmtExposure" : {"Not present" : 0, "No" : 0, "Mn" : 1, "Av": 2, "Gd" : 3},
"BsmtFinType1" : {"Not present" : 0, "Unf" : 1, "LwQ": 2, "Rec" : 3, "BLQ" : 4,
"ALQ" : 5, "GLQ" : 6},
"BsmtFinType2" : {"Not present" : 0, "Unf" : 1, "LwQ": 2, "Rec" : 3, "BLQ" : 4,
"ALQ" : 5, "GLQ" : 6},
"BsmtQual" : {"Not present" : 0, "Po" : 1, "Fa" : 2, "TA": 3, "Gd" : 4, "Ex" : 5},
"CentralAir" : {"N" : 0, "Y" : 1},
"ExterCond" : {"Po" : 1, "Fa" : 2, "TA": 3, "Gd": 4, "Ex" : 5},
"ExterQual" : {"Po" : 1, "Fa" : 2, "TA": 3, "Gd": 4, "Ex" : 5},
"FireplaceQu" : {"Not present" : 0, "Po" : 1, "Fa" : 2, "TA" : 3, "Gd" : 4, "Ex" : 5},
"Functional" : {"Sal" : 1, "Sev" : 2, "Maj2" : 3, "Maj1" : 4, "Mod": 5,
"Min2" : 6, "Min1" : 7, "Typ" : 8},
"GarageCond" : {"Not present" : 0, "Po" : 1, "Fa" : 2, "TA" : 3, "Gd" : 4, "Ex" : 5},
"GarageQual" : {"Not present" : 0, "Po" : 1, "Fa" : 2, "TA" : 3, "Gd" : 4, "Ex" : 5},
"GarageFinish" :{"Not present" : 0, "Unf" : 1, "RFn" : 2, "Fin" : 3},
"HeatingQC" : {"Po" : 1, "Fa" : 2, "TA" : 3, "Gd" : 4, "Ex" : 5},
"KitchenQual" : {"Po" : 1, "Fa" : 2, "TA" : 3, "Gd" : 4, "Ex" : 5},
"LandSlope" : {"Sev" : 1, "Mod" : 2, "Gtl" : 3},
"LotShape" : {"IR3" : 1, "IR2" : 2, "IR1" : 3, "Reg" : 4},
"PavedDrive" : {"N" : 0, "P" : 1, "Y" : 2},
"PoolQC" : {"Not present" : 0, "Fa" : 1, "TA" : 2, "Gd" : 3, "Ex" : 4},
"Street" : {"Grvl" : 1, "Pave" : 2},
"Utilities" : {"ELO" : 1, "NoSeWa" : 2, "NoSewr" : 3, "AllPub" : 4},
"Fence": {"Not present" : 0, "MnWw" : 1, "GdWo" : 2, "MnPrv" : 3, "GdPrv" : 4 }},
)
return df
<categorify> | def create_ResNet152_model() :
pretrained_model = tf.keras.applications.ResNet152V2(weights = 'imagenet', include_top = False, input_shape = [*IMAGE_SIZE, 3])
pretrained_model.trainable = True
model = tf.keras.Sequential([
pretrained_model,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation = 'softmax')
])
model.compile(optimizer = 'adam', loss = 'sparse_categorical_crossentropy', metrics = ['sparse_categorical_accuracy'])
return model | Petals to the Metal - Flower Classification on TPU |
10,378,659 | df_train = label_encode(df_train)
df_test = label_encode(df_test )<feature_engineering> | def create_MobileNetV2_model() :
pretrained_model = tf.keras.applications.MobileNetV2(weights = 'imagenet', include_top = False, input_shape = [*IMAGE_SIZE, 3])
pretrained_model.trainable = True
model = tf.keras.Sequential([
pretrained_model,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation = 'softmax')
])
model.compile(optimizer = 'adam', loss = 'sparse_categorical_crossentropy', metrics = ['sparse_categorical_accuracy'])
return model | Petals to the Metal - Flower Classification on TPU |
10,378,659 | df_train['SalePrice']=np.log(df_train['SalePrice'])
_ = sns.distplot(df_train["SalePrice"] )<define_variables> | def create_InceptionResNetV2_model() :
pretrained_model = tf.keras.applications.InceptionResNetV2(weights = 'imagenet', include_top = False, input_shape = [*IMAGE_SIZE, 3])
pretrained_model.trainable = True
model = tf.keras.Sequential([
pretrained_model,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation = 'softmax')
])
model.compile(optimizer = 'adam', loss = 'sparse_categorical_crossentropy', metrics = ['sparse_categorical_accuracy'])
return model | Petals to the Metal - Flower Classification on TPU |
10,378,659 | cat_list = [col for col in df_train.columns if df_train[col].dtypes == object]
num_list = [col for col in df_train.columns if df_train[col].dtypes != object]<merge> | no_of_models = 2
models = [0] * no_of_models
start_model = 0
end_model = 2
model_indx_0 = start_model
model_indx_1 = start_model + 1
| Petals to the Metal - Flower Classification on TPU |
10,378,659 | categorical_data = df_train[cat_list]
numerical_data = df_train[num_list]
df_train = categorical_data.join(numerical_data )<merge> | val_probabilities = [0] * no_of_models
test_probabilities = [0] * no_of_models
all_probabilities = [0] * no_of_models | Petals to the Metal - Flower Classification on TPU |
10,378,659 | num_list.remove('SalePrice')
cat_test = df_test[cat_list]
num_test = df_test[num_list]
df_test = cat_test.join(num_test )<prepare_x_and_y> | with strategy.scope() :
for j in range(no_of_models):
models[j] = create_EfficientNet_model()
models[0].summary() | Petals to the Metal - Flower Classification on TPU |
10,378,659 | X = df_train.drop('SalePrice', axis=1 ).values
y = df_train['SalePrice'].values
df_test_values = df_test.values<categorify> | def write_history(j):
history_dict = [0] * no_of_models
for i in range(j + 1):
if(historys[i] != 0):
history_dict[i] = historys[i].history
filename = './' + this_run_file_prefix + 'model_history_' + str(j)+ '.pkl'
pklfile = open(filename, 'ab')
pickle.dump(history_dict, pklfile)
pklfile.close() | Petals to the Metal - Flower Classification on TPU |
10,378,659 | def encode(X):
labelencoder = LabelEncoder()
for i in range(len(cat_list)) :
X[:,i] = labelencoder.fit_transform(X[:,i])
for i in range(len(cat_list)) :
onehotencoder = OneHotEncoder(categorical_features=[i])
X = onehotencoder.fit_transform(X ).toarray()
X = X[:,i:]
<categorify> | EPOCHS = 20
historys = [0] * no_of_models
finished_models = 0
for j in range(start_model, end_model):
start_training = datetime.now()
print(start_training)
time_from_start_program_tdelta = start_training - start_time
if time_from_start_program_tdelta > end_training_by_tdelta:
print(j, 'time limit for doing training over, get out')
break
print('LR_EXP_DECAY:', LR_EXP_DECAY, '.LR_MAX:', LR_MAX)
historys[j] = models[j].fit(get_training_dataset() , steps_per_epoch = STEPS_PER_EPOCH, epochs = EPOCHS, validation_data = get_validation_dataset() , callbacks = [lr_callback, early_stop])
write_history(j)
filename = this_run_file_prefix + 'models_' + str(j)+ '.h5'
models[j].save(filename)
gc.collect()
finished_models = j + 1
print(datetime.now())
| Petals to the Metal - Flower Classification on TPU |
10,378,659 | encode(X)
encode(df_test_values )<normalization> | cmdataset = get_validation_dataset(ordered = True)
images_ds = cmdataset.map(lambda image, label: image)
labels_ds = cmdataset.map(lambda image, label: label ).unbatch()
cm_correct_labels = next(iter(labels_ds.batch(NUM_VALIDATION_IMAGES)) ).numpy() | Petals to the Metal - Flower Classification on TPU |
10,378,659 | sc = StandardScaler()
X = sc.fit_transform(X)
df_test_values = sc.transform(df_test_values )<split> | test_ds = get_test_dataset(ordered = True)
test_images_ds = test_ds.map(lambda image, idnum: image)
| Petals to the Metal - Flower Classification on TPU |
10,378,659 | X_train,X_test,y_train,y_test = train_test_split(X,y,test_size=0.2 )<compute_test_metric> | dataset = get_validation_dataset()
dataset = dataset.unbatch().batch(20)
batch = iter(dataset)
images, labels = next(batch ) | Petals to the Metal - Flower Classification on TPU |
10,378,659 | def rmsle(y, y_pred):
assert len(y)== len(y_pred)
terms_to_sum = [(math.log(y_pred[i] + 1)- math.log(y[i] + 1)) ** 2.0 for i,pred in enumerate(y_pred)]
return(sum(terms_to_sum)*(1.0/len(y)))** 0.5<train_model> | print(datetime.now())
for j in range(start_model, end_model):
val_probabilities[j] = models[j].predict(images_ds)
test_probabilities[j] = models[j].predict(test_images_ds)
all_probabilities[j] = models[j].predict(images)
print(datetime.now())
| Petals to the Metal - Flower Classification on TPU |
10,378,659 | lm = LinearRegression()
lm.fit(X_train,y_train)
y_pred_reg = lm.predict(X_test )<feature_engineering> | for j in range(start_model, finished_models):
display_training_curves(historys[j].history['loss'], historys[j].history['val_loss'], 'loss', 211)
display_training_curves(historys[j].history['sparse_categorical_accuracy'], historys[j].history['val_sparse_categorical_accuracy'], 'accuracy', 212)
for j in range(start_model, finished_models):
print('model number:', j, ', Train Accuracy:', max(historys[j].history['sparse_categorical_accuracy']), ', Validation Accuracy:', max(historys[j].history['val_sparse_categorical_accuracy']))
for j in range(start_model, finished_models):
print('model number:', j, ', Train Loss:', min(historys[j].history['loss']), ', Validation Loss:', min(historys[j].history['val_loss'])) | Petals to the Metal - Flower Classification on TPU |
10,378,659 | lm.intercept_<compute_test_metric> | def getFitPrecisionRecall(correct_labels, predictions):
score = f1_score(correct_labels, predictions, labels = range(len(CLASSES)) , average = 'macro')
precision = precision_score(correct_labels, predictions, labels = range(len(CLASSES)) , average = 'macro')
recall = recall_score(correct_labels, predictions, labels = range(len(CLASSES)) , average = 'macro')
return score, precision, recall | Petals to the Metal - Flower Classification on TPU |
10,378,659 | rmsle(np.exp(y_test), np.exp(y_pred_reg))<train_model> | cmat = confusion_matrix(cm_correct_labels, cm_predictions, labels = range(len(CLASSES)))
score, precision, recall = getFitPrecisionRecall(cm_correct_labels, cm_predictions)
cmat =(cmat.T / cmat.sum(axis = -1)).T
display_confusion_matrix(cmat, score, precision, recall)
print('f1 score: {:.3f}, precision: {:.3f}, recall: {:.3f}'.format(score, precision, recall)) | Petals to the Metal - Flower Classification on TPU |
10,378,659 | lasso_model = Lasso()
lasso_model.fit(X_train,y_train )<predict_on_test> | def create_submission_file(filename, probabilities):
predictions = np.argmax(probabilities, axis = -1)
print('Generating submission file...', filename)
test_ids_ds = test_ds.map(lambda image, idnum: idnum ).unbatch()
test_ids = next(iter(test_ids_ds.batch(NUM_TEST_IMAGES)) ).numpy().astype('U')
np.savetxt(filename, np.rec.fromarrays([test_ids, predictions]), fmt = ['%s', '%d'], delimiter = ',', header = 'id,label', comments = '')
| Petals to the Metal - Flower Classification on TPU |
10,378,659 | y_pred_lasso = lasso_model.predict(X_test )<compute_test_metric> | probabilities = np.zeros(( test_probabilities[0].shape))
for j in range(no_of_models):
probabilities = probabilities + test_probabilities[j]
filename = this_run_file_prefix + 'submission.csv'
create_submission_file(filename, probabilities)
create_submission_file('submission.csv', probabilities ) | Petals to the Metal - Flower Classification on TPU |
10,378,659 | rmsle(np.exp(y_test), np.exp(y_pred_lasso))<train_on_grid> | def combine_two(correct_labels, probability_0, probability_1):
print('Start.', datetime.now())
alphas0_to_try = np.linspace(0, 1, 101)
best_score = -1
best_alpha0 = -1
best_alpha1 = -1
best_precision = -1
best_recall = -1
best_val_predictions = None
for alpha0 in alphas0_to_try:
alpha1 = 1.0 - alpha0
probabilities = alpha0 * probability_0 + alpha1 * probability_1
predictions = np.argmax(probabilities, axis = -1)
score, precision, recall = getFitPrecisionRecall(correct_labels, predictions)
if score > best_score:
best_alpha0 = alpha0
best_alpha1 = alpha1
best_score = score
best_precision = precision
best_recall = recall
best_val_predictions = predictions
return best_alpha0, best_alpha1, best_val_predictions, best_score, best_precision, best_recall | Petals to the Metal - Flower Classification on TPU |
10,378,659 | lcv = LassoCV()
lcv.fit(X_train,y_train )<set_options> | def combine_three(correct_labels, probability_0, probability_1, probability_2):
print('Start.', datetime.now())
alphas0_to_try = np.linspace(0, 1, 101)
alphas1_to_try = np.linspace(0, 1, 101)
best_score = -1
best_alpha0 = -1
best_alpha1 = -1
best_alpha2 = -1
best_precision = -1
best_recall = -1
best_val_predictions = None
for alpha0 in alphas0_to_try:
for alpha1 in alphas1_to_try:
if(alpha0 + alpha1)> 1.0:
break
alpha2 = 1.0 - alpha0 - alpha1
probabilities = alpha0 * probability_0 + alpha1 * probability_1 + alpha2 * probability_2
predictions = np.argmax(probabilities, axis = -1)
score, precision, recall = getFitPrecisionRecall(correct_labels, predictions)
if score > best_score:
best_alpha0 = alpha0
best_alpha1 = alpha1
best_alpha2 = alpha2
best_score = score
best_precision = precision
best_recall = recall
best_val_predictions = predictions
return best_alpha0, best_alpha1, best_alpha2, best_val_predictions, best_score, best_precision, best_recall | Petals to the Metal - Flower Classification on TPU |
10,378,659 | lcv.alpha_<predict_on_test> | def get_best_combination(no_models, cm_correct_labels, val_probabilities, test_probabilities):
best_fit_score = -10000.0
best_predictions = 0
choose_filename = ''
curr_predictions = np.argmax(val_probabilities[0], axis = -1)
score, precision, recall = getFitPrecisionRecall(cm_correct_labels, curr_predictions)
print('f1 score: {:.3f}, precision: {:.3f}, recall: {:.3f}'.format(score, precision, recall))
filename = this_run_file_prefix + 'submission_0.csv'
if best_fit_score < score:
best_fit_score = score
best_predictions = curr_predictions
choose_filename = filename
create_submission_file('./submission.csv', test_probabilities[0])
create_submission_file(filename, test_probabilities[0])
if no_models > 1:
curr_predictions = np.argmax(val_probabilities[1], axis = -1)
score, precision, recall = getFitPrecisionRecall(cm_correct_labels, curr_predictions)
print('f1 score: {:.3f}, precision: {:.3f}, recall: {:.3f}'.format(score, precision, recall))
filename = this_run_file_prefix + 'submission_1.csv'
if best_fit_score < score:
best_fit_score = score
best_predictions = curr_predictions
choose_filename = filename
create_submission_file('./submission.csv', test_probabilities[1])
create_submission_file(filename, test_probabilities[1])
if no_models > 2:
curr_predictions = np.argmax(val_probabilities[2], axis = -1)
score, precision, recall = getFitPrecisionRecall(cm_correct_labels, curr_predictions)
print('f1 score: {:.3f}, precision: {:.3f}, recall: {:.3f}'.format(score, precision, recall))
filename = this_run_file_prefix + 'submission_2.csv'
if best_fit_score < score:
best_fit_score = score
best_predictions = curr_predictions
choose_filename = filename
create_submission_file('./submission.csv', test_probabilities[2])
create_submission_file(filename, test_probabilities[2])
if no_models > 1:
best_alpha0, best_alpha1, best_val_predictions, best_score, best_precision, best_recall = combine_two(cm_correct_labels, val_probabilities[0], val_probabilities[1])
print('For indx', [0, 1], 'best_alpha0:', best_alpha0, 'best_alpha1:', best_alpha1, '.', datetime.now())
print('f1 score: {:.3f}, precision: {:.3f}, recall: {:.3f}'.format(best_score, best_precision, best_recall))
combined_probabilities = best_alpha0 * test_probabilities[0] + best_alpha1 * test_probabilities[1]
filename = this_run_file_prefix + 'submission_01.csv'
if best_fit_score < best_score:
best_fit_score = best_score
best_predictions = best_val_predictions
choose_filename = filename
create_submission_file('./submission.csv', combined_probabilities)
create_submission_file(filename, combined_probabilities)
if no_models > 2:
best_alpha0, best_alpha1, best_val_predictions, best_score, best_precision, best_recall = combine_two(cm_correct_labels, val_probabilities[0], val_probabilities[2])
print('For indx', [0, 2], 'best_alpha0:', best_alpha0, 'best_alpha1:', best_alpha1, '.', datetime.now())
print('f1 score: {:.3f}, precision: {:.3f}, recall: {:.3f}'.format(best_score, best_precision, best_recall))
combined_probabilities = best_alpha0 * test_probabilities[0] + best_alpha1 * test_probabilities[2]
filename = this_run_file_prefix + 'submission_02.csv'
if best_fit_score < best_score:
best_fit_score = best_score
best_predictions = best_val_predictions
choose_filename = filename
create_submission_file('./submission.csv', combined_probabilities)
create_submission_file(filename, combined_probabilities)
best_alpha0, best_alpha1, best_val_predictions, best_score, best_precision, best_recall = combine_two(cm_correct_labels, val_probabilities[1], val_probabilities[2])
print('For indx', [1, 2], 'best_alpha0:', best_alpha0, 'best_alpha1:', best_alpha1, '.', datetime.now())
print('f1 score: {:.3f}, precision: {:.3f}, recall: {:.3f}'.format(best_score, best_precision, best_recall))
combined_probabilities = best_alpha0 * test_probabilities[1] + best_alpha1 * test_probabilities[2]
filename = this_run_file_prefix + 'submission_12.csv'
if best_fit_score < best_score:
best_fit_score = best_score
best_predictions = best_val_predictions
choose_filename = filename
create_submission_file('./submission.csv', combined_probabilities)
create_submission_file(filename, combined_probabilities)
best_alpha0, best_alpha1, best_alpha2, best_val_predictions, best_score, best_precision, best_recall = combine_three(cm_correct_labels, val_probabilities[0], val_probabilities[1], val_probabilities[2])
print('For indx', [0, 1, 2], 'best_alpha0:', best_alpha0, 'best_alpha1:', best_alpha1, 'best_alpha2:', best_alpha2, '.', datetime.now())
print('f1 score: {:.3f}, precision: {:.3f}, recall: {:.3f}'.format(best_score, best_precision, best_recall))
combined_probabilities = best_alpha0 * test_probabilities[0] + best_alpha1 * test_probabilities[1] + best_alpha2 * test_probabilities[2]
filename = this_run_file_prefix + 'submission_012.csv'
if best_fit_score < best_score:
best_fit_score = best_score
best_predictions = best_val_predictions
choose_filename = filename
create_submission_file('./submission.csv', combined_probabilities)
create_submission_file(filename, combined_probabilities)
cmat = confusion_matrix(cm_correct_labels, best_predictions, labels = range(len(CLASSES)))
cmat =(cmat.T / cmat.sum(axis = -1)).T
display_confusion_matrix(cmat, best_fit_score, precision, recall)
print('Best score from all combination was', best_fit_score, '.For submission file used is', choose_filename)
return best_predictions
| Petals to the Metal - Flower Classification on TPU |
10,378,659 | y_pred_lassocv = lcv.predict(X_test )<compute_test_metric> | best_predictions = cm_predictions
if no_of_models > 1:
bp = get_best_combination(no_of_models, cm_correct_labels, val_probabilities, test_probabilities)
best_predictions = bp
| Petals to the Metal - Flower Classification on TPU |
10,378,659 | rmsle(np.exp(y_test),np.exp(y_pred_lassocv))<choose_model_class> | probabilities = np.zeros(( all_probabilities[0].shape))
for j in range(no_of_models):
probabilities = probabilities + all_probabilities[j]
predictions = np.argmax(probabilities, axis =-1)
display_batch_of_images(( images, labels), predictions ) | Petals to the Metal - Flower Classification on TPU |
10,378,659 | model = Lasso(lcv.alpha_)
model.fit(X,y)
y_pred = model.predict(df_test_values)
predictions = np.exp(y_pred )<save_to_csv> | val_probs = [cm_correct_labels, cm_predictions, val_probabilities[0], val_probabilities[1], test_probabilities[0], test_probabilities[1]]
filename = this_run_file_prefix + 'tests_vals_01.pkl'
pklfile = open(filename, 'ab')
pickle.dump(val_probs, pklfile)
pklfile.close() | Petals to the Metal - Flower Classification on TPU |
10,378,659 | result=pd.DataFrame({'Id':df_test.index, 'SalePrice':predictions})
result.to_csv('submission.csv', index=False )<set_options> | use_correct_labels = cm_correct_labels
use_val_predictions = best_predictions | Petals to the Metal - Flower Classification on TPU |
10,378,659 | warnings.filterwarnings('ignore' )<compute_test_metric> | correct_labels_cnt = 0
incorrect_labels_cnt = 0
correct_labels = []
incorrect_labels = []
vals_actual_true = {}
vals_tp = {}
vals_fn = {}
vals_fp = {}
for i in range(len(CLASSES)) :
vals_actual_true[i] = 0
vals_tp[i] = 0
vals_fn[i] = 0
vals_fp[i] = 0
for i in range(len(use_correct_labels)) :
correct_label = use_correct_labels[i]
predict_label = use_val_predictions[i]
vals_actual_true[correct_label] = vals_actual_true[correct_label] + 1
if use_val_predictions[i] != use_correct_labels[i]:
incorrect_labels_cnt = incorrect_labels_cnt + 1
incorrect_labels.append(i)
vals_fn[correct_label] = vals_fn[correct_label] + 1
vals_fp[predict_label] = vals_fp[predict_label] + 1
else:
correct_labels_cnt = correct_labels_cnt + 1
correct_labels.append(i)
vals_tp[correct_label] = vals_tp[correct_label] + 1
print('Number of correct_labels is {}, incorrect_labels is {}'.format(correct_labels_cnt, incorrect_labels_cnt))
print('Incorrect labels', incorrect_labels)
| Petals to the Metal - Flower Classification on TPU |
10,378,659 | def rmse_cv(model, X, y):
rmse = np.sqrt(-cross_val_score(model, X_train, y_train, scoring="neg_mean_squared_error", cv=kfolds))
return(rmse)
def rmsle(y, y_pred):
return(np.sqrt(metrics.mean_squared_error(y, y_pred)))
def CVscore(model, X, y):
result = cross_val_score(model, X, y, cv=kfold)
return result.mean()
def blend_models_predict(X, models):
blend = []
for model in models:
blend.append(1/len(models)*model.predict(X))
return sum(blend )<load_from_csv> | def display_my_batch_of_images(databatch, rows = 0, cols = 0, predictions=None):
images, labels = databatch
if labels is None:
labels = [None for _ in enumerate(images)]
if rows == 0 or cols == 0:
rows = int(math.sqrt(len(images)))
cols =(len(images)+ rows - 1)//rows
print('Total number of images is {}, rows is {}, cols is {}'.format(len(images), rows, cols))
FIGSIZE = 20.0
SPACING = 0.1
subplot=(rows,cols,1)
if rows < cols:
plt.figure(figsize=(FIGSIZE,FIGSIZE/cols*rows))
else:
plt.figure(figsize=(FIGSIZE/rows*cols,FIGSIZE))
for i,(image, label)in enumerate(zip(images[:rows*cols], labels[:rows*cols])) :
title = '' if label is None else CLASSES[label]
correct = True
if predictions is not None:
title, correct = title_from_label_and_target(predictions[i], label)
dynamic_titlesize = FIGSIZE*SPACING/max(rows,cols)*40+3
subplot = display_one_flower(image, title, subplot, not correct, titlesize=dynamic_titlesize)
plt.tight_layout()
if label is None and predictions is None:
plt.subplots_adjust(wspace=0, hspace=0)
else:
plt.subplots_adjust(wspace=SPACING, hspace=SPACING)
plt.show() | Petals to the Metal - Flower Classification on TPU |
10,378,659 | df_train = pd.read_csv('.. /input/train.csv',index_col='Id')
df_test = pd.read_csv('.. /input/test.csv',index_col='Id' )<drop_column> | disp_labels = []
disp_predictions = []
for i in range(54):
if i >= incorrect_labels_cnt:
break
id = incorrect_labels[i]
disp_labels.append(use_correct_labels[id])
disp_predictions.append(use_val_predictions[id])
print(disp_labels)
print(disp_predictions ) | Petals to the Metal - Flower Classification on TPU |
10,378,659 | <define_variables><EOS> | val_ids = list(range(len(use_correct_labels)))
filename = this_run_file_prefix + 'validation_results.csv'
np.savetxt(filename, np.rec.fromarrays([val_ids, use_correct_labels, use_val_predictions]), fmt = ['%d', '%d', '%d'], delimiter = ',', header = 'id,correct_label,predicted_label', comments = '')
cls_ids = list(range(len(CLASSES)))
filename = this_run_file_prefix + 'validation_statistics.csv'
np.savetxt(filename, np.rec.fromarrays([cls_ids, list(vals_actual_true.values()), list(vals_tp.values()), list(vals_fn.values()), list(vals_fp.values())]), fmt = ['%d', '%d', '%d', '%d', '%d'], delimiter = ',', header = 'cls_id,actual_true,true_positive,false_negative,false_positive', comments = '' ) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | <SOS> metric: macrofscore Kaggle data source: petals-to-the-metal-flower-classification-on-tpu<prepare_x_and_y> | print("TF version " + tf.__version__)
AUTO = tf.data.experimental.AUTOTUNE | Petals to the Metal - Flower Classification on TPU |
10,218,760 | y = df_train['SalePrice']<concatenate> | try:
tpu = tf.distribute.cluster_resolver.TPUClusterResolver()
print('Running on TPU ', tpu.master())
except ValueError:
tpu = None
if tpu:
tf.config.experimental_connect_to_cluster(tpu)
tf.tpu.experimental.initialize_tpu_system(tpu)
strategy = tf.distribute.experimental.TPUStrategy(tpu)
else:
strategy = tf.distribute.get_strategy()
print("REPLICAS: ", strategy.num_replicas_in_sync ) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | df_all = pd.concat([df_train, df_test], sort=False )<categorify> | IMAGE_SIZE = [512, 512]
EPOCHS = 40
BATCH_SIZE = 16 * strategy.num_replicas_in_sync
SEED = 752
SKIP_VALIDATION = False
TTA_NUM = 5
random.seed(SEED)
np.random.seed(SEED)
tf.random.set_seed(SEED ) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | Cat_toNone =('PoolQC','Fence','MiscFeature','Alley','FireplaceQu','GarageType','GarageFinish',
'GarageQual','GarageCond','BsmtQual','BsmtCond','BsmtExposure','BsmtFinType1',
'BsmtFinType2','MasVnrType','MSSubClass')
for c in Cat_toNone:
df_all[c] = df_all[c].fillna('None' )<categorify> | np.set_printoptions(threshold=15, linewidth=80)
def batch_to_numpy_images_and_labels(data):
images, labels = data
numpy_images = images.numpy()
numpy_labels = labels.numpy()
if numpy_labels.dtype == object:
numpy_labels = [None for _ in enumerate(numpy_images)]
return numpy_images, numpy_labels
def title_from_label_and_target(label, correct_label):
if correct_label is None:
return CLASSES[label], True
correct =(label == correct_label)
return "{} [{}{}{}]".format(CLASSES[label], 'OK' if correct else 'NO', u"\u2192" if not correct else '',
CLASSES[correct_label] if not correct else ''), correct
def display_one_flower(image, title, subplot, red=False, titlesize=16):
plt.subplot(*subplot)
plt.axis('off')
plt.imshow(image)
if len(title)> 0:
plt.title(title, fontsize=int(titlesize)if not red else int(titlesize/1.2), color='red' if red else 'black', fontdict={'verticalalignment':'center'}, pad=int(titlesize/1.5))
return(subplot[0], subplot[1], subplot[2]+1)
def display_batch_of_images(databatch, predictions=None):
images, labels = batch_to_numpy_images_and_labels(databatch)
if labels is None:
labels = [None for _ in enumerate(images)]
rows = int(math.sqrt(len(images)))
cols = len(images)//rows
FIGSIZE = 13.0
SPACING = 0.1
subplot=(rows,cols,1)
if rows < cols:
plt.figure(figsize=(FIGSIZE,FIGSIZE/cols*rows))
else:
plt.figure(figsize=(FIGSIZE/rows*cols,FIGSIZE))
for i,(image, label)in enumerate(zip(images[:rows*cols], labels[:rows*cols])) :
title = '' if label is None else CLASSES[label]
correct = True
if predictions is not None:
title, correct = title_from_label_and_target(predictions[i], label)
dynamic_titlesize = FIGSIZE*SPACING/max(rows,cols)*40+3
subplot = display_one_flower(image, title, subplot, not correct, titlesize=dynamic_titlesize)
plt.tight_layout()
if label is None and predictions is None:
plt.subplots_adjust(wspace=0, hspace=0)
else:
plt.subplots_adjust(wspace=SPACING, hspace=SPACING)
plt.show()
def display_confusion_matrix(cmat, score, precision, recall):
plt.figure(figsize=(15,15))
ax = plt.gca()
ax.matshow(cmat, cmap='Reds')
ax.set_xticks(range(len(CLASSES)))
ax.set_xticklabels(CLASSES, fontdict={'fontsize': 7})
plt.setp(ax.get_xticklabels() , rotation=45, ha="left", rotation_mode="anchor")
ax.set_yticks(range(len(CLASSES)))
ax.set_yticklabels(CLASSES, fontdict={'fontsize': 7})
plt.setp(ax.get_yticklabels() , rotation=45, ha="right", rotation_mode="anchor")
titlestring = ""
if score is not None:
titlestring += 'f1 = {:.3f} '.format(score)
if precision is not None:
titlestring += '
precision = {:.3f} '.format(precision)
if recall is not None:
titlestring += '
recall = {:.3f} '.format(recall)
if len(titlestring)> 0:
ax.text(101, 1, titlestring, fontdict={'fontsize': 18, 'horizontalalignment':'right', 'verticalalignment':'top', 'color':'
plt.show()
def display_training_curves(training, validation, title, subplot):
if subplot%10==1:
plt.subplots(figsize=(10,10), facecolor='
plt.tight_layout()
ax = plt.subplot(subplot)
ax.set_facecolor('
ax.plot(training)
ax.plot(validation)
ax.set_title('model '+ title)
ax.set_ylabel(title)
ax.set_xlabel('epoch')
ax.legend(['train', 'valid.'] ) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | Cat_toMode =('MSZoning','Electrical','KitchenQual','Exterior1st','Exterior2nd',
'Utilities','SaleType','Functional')
for c in Cat_toMode:
df_all[c] = df_all[c].fillna(df_all[c].mode() [0] )<data_type_conversions> | def random_blockout(img, sl=0.1, sh=0.2, rl=0.4):
p=random.random()
if p>=0.25:
w, h, c = IMAGE_SIZE[0], IMAGE_SIZE[1], 3
origin_area = tf.cast(h*w, tf.float32)
e_size_l = tf.cast(tf.round(tf.sqrt(origin_area * sl * rl)) , tf.int32)
e_size_h = tf.cast(tf.round(tf.sqrt(origin_area * sh / rl)) , tf.int32)
e_height_h = tf.minimum(e_size_h, h)
e_width_h = tf.minimum(e_size_h, w)
erase_height = tf.random.uniform(shape=[], minval=e_size_l, maxval=e_height_h, dtype=tf.int32)
erase_width = tf.random.uniform(shape=[], minval=e_size_l, maxval=e_width_h, dtype=tf.int32)
erase_area = tf.zeros(shape=[erase_height, erase_width, c])
erase_area = tf.cast(erase_area, tf.uint8)
pad_h = h - erase_height
pad_top = tf.random.uniform(shape=[], minval=0, maxval=pad_h, dtype=tf.int32)
pad_bottom = pad_h - pad_top
pad_w = w - erase_width
pad_left = tf.random.uniform(shape=[], minval=0, maxval=pad_w, dtype=tf.int32)
pad_right = pad_w - pad_left
erase_mask = tf.pad([erase_area], [[0,0],[pad_top, pad_bottom], [pad_left, pad_right], [0,0]], constant_values=1)
erase_mask = tf.squeeze(erase_mask, axis=0)
erased_img = tf.multiply(tf.cast(img,tf.float32), tf.cast(erase_mask, tf.float32))
return tf.cast(erased_img, img.dtype)
else:
return tf.cast(img, img.dtype ) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | Num_toZero =('GarageArea','GarageCars','BsmtFinSF1','BsmtFinSF2','BsmtUnfSF','TotalBsmtSF',
'BsmtFullBath','BsmtHalfBath','GarageYrBlt','MasVnrArea')
for c in Num_toZero:
df_all[c] = df_all[c].fillna(0 )<feature_engineering> | def decode_image(image_data):
image = tf.image.decode_jpeg(image_data, channels=3)
image = tf.cast(image, tf.float32)/ 255.0
image = tf.reshape(image, [*IMAGE_SIZE, 3])
return image
def read_labeled_tfrecord(example):
LABELED_TFREC_FORMAT = {
"image": tf.io.FixedLenFeature([], tf.string),
"class": tf.io.FixedLenFeature([], tf.int64),
}
example = tf.io.parse_single_example(example, LABELED_TFREC_FORMAT)
image = decode_image(example['image'])
label = tf.cast(example['class'], tf.int32)
return image, label
def read_unlabeled_tfrecord(example):
UNLABELED_TFREC_FORMAT = {
"image": tf.io.FixedLenFeature([], tf.string),
"id": tf.io.FixedLenFeature([], tf.string),
}
example = tf.io.parse_single_example(example, UNLABELED_TFREC_FORMAT)
image = decode_image(example['image'])
idnum = example['id']
return image, idnum
def load_dataset(filenames, labeled=True, ordered=False):
ignore_order = tf.data.Options()
if not ordered:
ignore_order.experimental_deterministic = False
dataset = tf.data.TFRecordDataset(filenames, num_parallel_reads=AUTO)
dataset = dataset.with_options(ignore_order)
dataset = dataset.map(read_labeled_tfrecord if labeled else read_unlabeled_tfrecord, num_parallel_calls=AUTO)
return dataset
def data_augment(image, label):
image = tf.image.random_flip_left_right(image, seed=SEED)
image = random_blockout(image)
return image, label
def get_training_dataset() :
dataset = load_dataset(TRAINING_FILENAMES, labeled=True)
dataset = dataset.map(data_augment, num_parallel_calls=AUTO)
dataset = dataset.repeat()
dataset = dataset.shuffle(2048)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.prefetch(AUTO)
return dataset
def get_validation_dataset(ordered=False):
dataset = load_dataset(VALIDATION_FILENAMES, labeled=True, ordered=ordered)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.cache()
dataset = dataset.prefetch(AUTO)
return dataset
def get_test_dataset(ordered=False):
dataset = load_dataset(TEST_FILENAMES, labeled=False, ordered=ordered)
dataset = dataset.batch(BATCH_SIZE)
dataset = dataset.prefetch(AUTO)
return dataset
def count_data_items(filenames):
n = [int(re.compile(r"-([0-9]*)\." ).search(filename ).group(1)) for filename in filenames]
return np.sum(n)
NUM_TRAINING_IMAGES = count_data_items(TRAINING_FILENAMES)
NUM_VALIDATION_IMAGES = count_data_items(VALIDATION_FILENAMES)
NUM_TEST_IMAGES = count_data_items(TEST_FILENAMES)
STEPS_PER_EPOCH = NUM_TRAINING_IMAGES // BATCH_SIZE
print('Dataset: {} training images, {} validation images, {} unlabeled test images'.format(NUM_TRAINING_IMAGES, NUM_VALIDATION_IMAGES, NUM_TEST_IMAGES)) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | df_all["LotFrontage"] = df_all.groupby("Neighborhood")["LotFrontage"].transform(lambda x: x.fillna(x.median()))<data_type_conversions> | if SKIP_VALIDATION:
TRAINING_FILENAMES = TRAINING_FILENAMES + VALIDATION_FILENAMES | Petals to the Metal - Flower Classification on TPU |
10,218,760 | df_all['MSSubClass'] = df_all['MSSubClass'].astype(str )<feature_engineering> | print("Training data shapes:")
for image, label in get_training_dataset().take(3):
print(image.numpy().shape, label.numpy().shape)
print("Training data label examples:", label.numpy())
print("Validation data shapes:")
for image, label in get_validation_dataset().take(3):
print(image.numpy().shape, label.numpy().shape)
print("Validation data label examples:", label.numpy())
print("Test data shapes:")
for image, idnum in get_test_dataset().take(3):
print(image.numpy().shape, idnum.numpy().shape)
print("Test data IDs:", idnum.numpy().astype('U')) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | skewed_feats = df_all[numeric].apply(lambda x: skew(x.dropna()))
skewed_feats = skewed_feats[skewed_feats > 0.75]
skewed_feats = skewed_feats.index
df_all[skewed_feats] = np.log1p(df_all[skewed_feats])
<categorify> | training_dataset = get_training_dataset()
training_dataset = training_dataset.unbatch().batch(20)
train_batch = iter(training_dataset ) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | df_all = df_all.replace({"Alley" : {"None" : 0, "Grvl" : 1, "Pave" : 2},
"BsmtCond" : {"None" : 0, "Po" : 1, "Fa" : 2, "TA" : 3, "Gd" : 4, "Ex" : 5},
"BsmtExposure" : {"None" : 0, "Mn" : 1, "Av": 2, "Gd" : 3},
"BsmtFinType1" : {"None" : 0, "Unf" : 1, "LwQ": 2, "Rec" : 3, "BLQ" : 4,
"ALQ" : 5, "GLQ" : 6},
"BsmtFinType2" : {"None" : 0, "Unf" : 1, "LwQ": 2, "Rec" : 3, "BLQ" : 4,
"ALQ" : 5, "GLQ" : 6},
"BsmtQual" : {"None" : 0, "Po" : 1, "Fa" : 2, "TA": 3, "Gd" : 4, "Ex" : 5},
"ExterCond" : {"Po" : 1, "Fa" : 2, "TA": 3, "Gd": 4, "Ex" : 5},
"ExterQual" : {"Po" : 1, "Fa" : 2, "TA": 3, "Gd": 4, "Ex" : 5},
"FireplaceQu" : {"None" : 0, "Po" : 1, "Fa" : 2, "TA" : 3, "Gd" : 4, "Ex" : 5},
"Functional" : {"Sal" : 1, "Sev" : 2, "Maj2" : 3, "Maj1" : 4, "Mod": 5,
"Min2" : 6, "Min1" : 7, "Typ" : 8},
"GarageCond" : {"None" : 0, "Po" : 1, "Fa" : 2, "TA" : 3, "Gd" : 4, "Ex" : 5},
"GarageQual" : {"None" : 0, "Po" : 1, "Fa" : 2, "TA" : 3, "Gd" : 4, "Ex" : 5},
"HeatingQC" : {"Po" : 1, "Fa" : 2, "TA" : 3, "Gd" : 4, "Ex" : 5},
"KitchenQual" : {"Po" : 1, "Fa" : 2, "TA" : 3, "Gd" : 4, "Ex" : 5},
"LandSlope" : {"Sev" : 1, "Mod" : 2, "Gtl" : 3},
"LotShape" : {"IR3" : 1, "IR2" : 2, "IR1" : 3, "Reg" : 4},
"PavedDrive" : {"N" : 0, "P" : 1, "Y" : 2},
"PoolQC" : {"None" : 0, "Fa" : 1, "TA" : 2, "Gd" : 3, "Ex" : 4},
"Street" : {"Grvl" : 1, "Pave" : 2},
"Utilities" : {"ELO" : 1, "NoSeWa" : 2, "NoSewr" : 3, "AllPub" : 4}}
)<feature_engineering> | display_batch_of_images(next(train_batch)) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | df_all['Bathrooms'] = df_all['FullBath'] +(df_all['HalfBath']*0.5)+ df_all['BsmtFullBath'] +(df_all['BsmtHalfBath']*0.5)
df_all['FireplaceScore'] = df_all['Fireplaces'] * df_all['FireplaceQu']
df_all['GarageScore'] = df_all['GarageCars'] * df_all['GarageQual'] * df_all['GarageArea']
df_all['HouseAge'] = df_all['YrSold'] - df_all['YearBuilt']
df_all['TotalLivingSF'] = df_all['GrLivArea'] + df_all['TotalBsmtSF'] - df_all['LowQualFinSF']<drop_column> | test_dataset = get_test_dataset()
test_dataset = test_dataset.unbatch().batch(20)
test_batch = iter(test_dataset ) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | df_all = df_all.drop(['FullBath', 'HalfBath', 'BsmtFullBath', 'BsmtHalfBath',
'GarageCars','GarageArea','GarageQual','GarageCond',
'FireplaceQu','Fireplaces',
'PoolArea',
'Utilities', 'Street',
'GarageYrBlt'
], axis = 1 )<categorify> | display_batch_of_images(next(test_batch)) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | df_all = pd.get_dummies(df_all, drop_first=True )<drop_column> | LR_START = 0.00001
LR_MAX = 0.00005 * strategy.num_replicas_in_sync
LR_MIN = 0.00001
LR_RAMPUP_EPOCHS = 5
LR_SUSTAIN_EPOCHS = 0
LR_EXP_DECAY =.8
def lrfn(epoch):
if epoch < LR_RAMPUP_EPOCHS:
lr =(LR_MAX - LR_START)/ LR_RAMPUP_EPOCHS * epoch + LR_START
elif epoch < LR_RAMPUP_EPOCHS + LR_SUSTAIN_EPOCHS:
lr = LR_MAX
else:
lr =(LR_MAX - LR_MIN)* LR_EXP_DECAY**(epoch - LR_RAMPUP_EPOCHS - LR_SUSTAIN_EPOCHS)+ LR_MIN
return lr
lr_callback = tf.keras.callbacks.LearningRateScheduler(lrfn, verbose = True)
rng = [i for i in range(25 if EPOCHS<25 else EPOCHS)]
y = [lrfn(x)for x in rng]
plt.plot(rng, y)
print("Learning rate schedule: {:.3g} to {:.3g} to {:.3g}".format(y[0], max(y), y[-1])) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | df_all = df_all.drop(['SalePrice'], axis = 1)
train = df_all[:len_train]
test = df_all[len_train:]<split> | !pip install -q efficientnet
| Petals to the Metal - Flower Classification on TPU |
10,218,760 | X_train, X_test, y_train, y_test = train_test_split(train, y, test_size = 0.3, random_state = 0 )<choose_model_class> | with strategy.scope() :
enet = efn.EfficientNetB7(input_shape=[*IMAGE_SIZE, 3], weights='noisy-student', include_top=False)
enet.trainable = True
model1 = tf.keras.Sequential([
enet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation='softmax')
])
model1.compile(
optimizer=tf.keras.optimizers.Adam() ,
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
model1.summary() | Petals to the Metal - Flower Classification on TPU |
10,218,760 | regressor = LinearRegression()
regressor.fit(X_train, y_train)
train_lm = regressor.predict(X_train)
test_lm = regressor.predict(X_test)
ridge = Ridge(alpha = 10)
ridge.fit(X_train, y_train)
ridge_train_lm = ridge.predict(X_train)
ridge_test_lm = ridge.predict(X_test)
lasso = Lasso(max_iter=500,alpha = 0.001)
lasso.fit(X_train, y_train)
lasso_train_lm = lasso.predict(X_train)
lasso_test_lm = lasso.predict(X_test)
en = ElasticNet(max_iter=500,alpha = 0.001)
en.fit(X_train, y_train)
en_train_lm = en.predict(X_train)
en_test_lm = en.predict(X_test)
dtr = DecisionTreeRegressor(max_depth=5)
dtr.fit(X_train, y_train)
train_dtr = dtr.predict(X_train)
test_dtr = dtr.predict(X_test)
rf = RandomForestRegressor(random_state = 0)
rf.fit(X_train, y_train)
train_rf = rf.predict(X_train)
test_rf = rf.predict(X_test )<compute_train_metric> | if not SKIP_VALIDATION:
history1 = model1.fit(get_training_dataset() ,
steps_per_epoch=STEPS_PER_EPOCH,
epochs=EPOCHS,
validation_data=get_validation_dataset() ,
callbacks = [lr_callback])
else:
history1 = model1.fit(get_training_dataset() ,
steps_per_epoch=STEPS_PER_EPOCH,
epochs=EPOCHS,
callbacks = [lr_callback] ) | Petals to the Metal - Flower Classification on TPU |
10,218,760 | scorer = metrics.make_scorer(metrics.mean_squared_error, greater_is_better = False)
kfolds = KFold(n_splits=10, shuffle=True, random_state=42 )<compute_train_metric> | with strategy.scope() :
densenet = tf.keras.applications.DenseNet201(input_shape=[*IMAGE_SIZE, 3], weights='imagenet', include_top=False)
densenet.trainable = True
model2 = tf.keras.Sequential([
densenet,
tf.keras.layers.GlobalAveragePooling2D() ,
tf.keras.layers.Dense(len(CLASSES), activation='softmax')
])
model2.compile(
optimizer=tf.keras.optimizers.Adam() ,
loss = 'sparse_categorical_crossentropy',
metrics=['sparse_categorical_accuracy']
)
model2.summary() | Petals to the Metal - Flower Classification on TPU |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.