path
stringlengths 13
17
| screenshot_names
sequencelengths 1
873
| code
stringlengths 0
40.4k
| cell_type
stringclasses 1
value |
|---|---|---|---|
17134452/cell_11 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
import advertools as adv
import pandas as pd
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key
clubs = pd.read_csv('../input/clubs.csv')
serp_clubs = pd.read_csv('../input/serp_clubs.csv')
serp_clubs['totalResults'] = serp_clubs['totalResults'].astype('int')
serp_clubs['queryTime'] = pd.to_datetime(serp_clubs['queryTime'])
print(serp_clubs.shape)
serp_clubs.head(2) | code |
17134452/cell_19 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
import advertools as adv
import pandas as pd
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key
clubs = pd.read_csv('../input/clubs.csv')
serp_clubs = pd.read_csv('../input/serp_clubs.csv')
serp_clubs['totalResults'] = serp_clubs['totalResults'].astype('int')
serp_clubs['queryTime'] = pd.to_datetime(serp_clubs['queryTime'])
serp_clubs.drop_duplicates(['searchTerms']).groupby('searchTerms', as_index=False).agg({'totalResults': 'sum'}).sort_values('totalResults', ascending=False).reset_index(drop=True).head(15).style.format({'totalResults': '{:,}'}) | code |
17134452/cell_1 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
import advertools as adv
import pandas as pd
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__ | code |
17134452/cell_7 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
import advertools as adv
import pandas as pd
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key
clubs = pd.read_csv('../input/clubs.csv')
top_countries = clubs.groupby('Country').agg({'Total': 'sum'}).sort_values('Total', ascending=False).reset_index().head(10)
top_countries
clubs.groupby(['Country']).agg({'Club': 'count', 'Total': 'sum'}).sort_values('Club', ascending=False).reset_index().head(9).set_axis(['country', 'num_clubs', 'total_wins'], axis=1, inplace=False).assign(wins_per_club=lambda df: df['total_wins'].div(df['num_clubs'])).style.background_gradient(high=0.2)
clubs_list = clubs['Club'].str.lower().tolist()
clubs_list[:10] | code |
17134452/cell_18 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
import advertools as adv
import pandas as pd
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key
clubs = pd.read_csv('../input/clubs.csv')
serp_clubs = pd.read_csv('../input/serp_clubs.csv')
serp_clubs['totalResults'] = serp_clubs['totalResults'].astype('int')
serp_clubs['queryTime'] = pd.to_datetime(serp_clubs['queryTime'])
adv.word_frequency(serp_clubs['snippet'].fillna(''), phrase_len=2, rm_words=adv.stopwords['english'].union(['-', '|', ' ', '', 'de'])).head(20) | code |
17134452/cell_8 | [
"application_vnd.jupyter.stderr_output_1.png"
] | lang_football = {'en': 'football', 'fr': 'football', 'de': 'fußball', 'es': 'fútbol', 'it': 'calcio', 'pt-BR': 'futebol', 'nl': 'voetbal'}
lang_football | code |
17134452/cell_15 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
import advertools as adv
import pandas as pd
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key
clubs = pd.read_csv('../input/clubs.csv')
serp_clubs = pd.read_csv('../input/serp_clubs.csv')
serp_clubs['totalResults'] = serp_clubs['totalResults'].astype('int')
serp_clubs['queryTime'] = pd.to_datetime(serp_clubs['queryTime'])
(serp_clubs['snippet'].isna().sum(), serp_clubs['title'].isna().sum()) | code |
17134452/cell_16 | [
"text_plain_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
import advertools as adv
import pandas as pd
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key
clubs = pd.read_csv('../input/clubs.csv')
serp_clubs = pd.read_csv('../input/serp_clubs.csv')
serp_clubs['totalResults'] = serp_clubs['totalResults'].astype('int')
serp_clubs['queryTime'] = pd.to_datetime(serp_clubs['queryTime'])
adv.word_frequency(serp_clubs['snippet'].fillna(''), rm_words=adv.stopwords['english'].union(['-', '|', ' ', '·', '', 'de'])).head(15) | code |
17134452/cell_3 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
import advertools as adv
import pandas as pd
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key | code |
17134452/cell_17 | [
"text_plain_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
import advertools as adv
import pandas as pd
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key
clubs = pd.read_csv('../input/clubs.csv')
serp_clubs = pd.read_csv('../input/serp_clubs.csv')
serp_clubs['totalResults'] = serp_clubs['totalResults'].astype('int')
serp_clubs['queryTime'] = pd.to_datetime(serp_clubs['queryTime'])
adv.word_frequency(serp_clubs[serp_clubs['hl'] == 'en']['snippet'].fillna(''), rm_words=adv.stopwords['english'].union(['-', '|', ' ', '·', '', 'de'])).head(15) | code |
17134452/cell_24 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
from plotly.tools import make_subplots
import advertools as adv
import pandas as pd
import plotly.graph_objs as go
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key
clubs = pd.read_csv('../input/clubs.csv')
top_countries = clubs.groupby('Country').agg({'Total': 'sum'}).sort_values('Total', ascending=False).reset_index().head(10)
top_countries
clubs.groupby(['Country']).agg({'Club': 'count', 'Total': 'sum'}).sort_values('Club', ascending=False).reset_index().head(9).set_axis(['country', 'num_clubs', 'total_wins'], axis=1, inplace=False).assign(wins_per_club=lambda df: df['total_wins'].div(df['num_clubs'])).style.background_gradient(high=0.2)
lang_football = {'en': 'football', 'fr': 'football', 'de': 'fußball', 'es': 'fútbol', 'it': 'calcio', 'pt-BR': 'futebol', 'nl': 'voetbal'}
lang_football
serp_clubs = pd.read_csv('../input/serp_clubs.csv')
serp_clubs['totalResults'] = serp_clubs['totalResults'].astype('int')
serp_clubs['queryTime'] = pd.to_datetime(serp_clubs['queryTime'])
club_country = {club.lower(): country.lower() for club, country in zip(clubs['Club'], clubs['Country'])}
football_multi = '|'.join([' ' + football for football in lang_football.values()])
serp_clubs['country'] = [club_country[club].title()
for club in serp_clubs['searchTerms'].str.replace(football_multi, '')]
serp_clubs[['searchTerms', 'country']].sample(10)
serp_clubs.drop_duplicates(['searchTerms']).groupby('searchTerms', as_index=False).agg({'totalResults': 'sum'}).sort_values('totalResults', ascending=False).reset_index(drop=True).head(15).style.format({'totalResults': '{:,}'})
hl_domain_appearances = serp_clubs.groupby(['hl', 'displayLink']).agg({'rank': 'count'}).reset_index().sort_values(['hl', 'rank'], ascending=False).rename(columns={'rank': 'search_appearances'})
hl_domain_appearances.groupby(['hl']).head(5)
fig = make_subplots(1, 7, print_grid=False, shared_yaxes=True)
for i, lang in enumerate(serp_clubs['hl'].unique()[:7]):
df = serp_clubs[serp_clubs['hl']==lang]
fig.append_trace(go.Bar(y=df['displayLink'].value_counts().values[:8],
x=df['displayLink'].value_counts().index.str.replace('www.', '')[:8],
name=lang,
orientation='v'), row=1, col=i+1)
fig.layout.margin = {'b': 150, 'r': 30}
fig.layout.legend.orientation = 'h'
fig.layout.legend.y = -0.5
fig.layout.legend.x = 0.15
fig.layout.title = 'Top Domains by Language of Search'
fig.layout.yaxis.title = 'Number of Appearances on SERPs'
fig.layout.plot_bgcolor = '#eeeeee'
fig.layout.paper_bgcolor = '#eeeeee'
iplot(fig)
fig = make_subplots(1, 7, shared_yaxes=True, print_grid=False)
for i, country in enumerate(serp_clubs['country'].unique()[:7]):
if country in top_countries['Country'][:7].values:
df = serp_clubs[serp_clubs['country']==country]
fig.append_trace(go.Bar(y=df['displayLink'].value_counts().values[:8],
x=df['displayLink'].value_counts().index.str.replace('www.', '')[:8],
name=country,
orientation='v'), row=1, col=i+1)
fig.layout.margin = {'b': 150, 'r': 0}
fig.layout.legend.orientation = 'h'
fig.layout.legend.y = -0.5
fig.layout.legend.x = 0.15
fig.layout.title = 'Top Domains by Country of Club'
fig.layout.yaxis.title = 'Number of Appearances on SERPs'
fig.layout.plot_bgcolor = '#eeeeee'
fig.layout.paper_bgcolor = '#eeeeee'
iplot(fig)
def plot_serps(df, opacity=0.1, num_domains=10, width=None, height=700):
"""
df: a DataFrame resulting from running advertools.serp_goog
opacity: the opacity of the markers [0, 1]
num_domains: how many domains to plot
"""
top_domains = df['displayLink'].value_counts()[:num_domains].index.tolist()
top_df = df[df['displayLink'].isin(top_domains)]
top_df_counts_means = top_df.groupby('displayLink', as_index=False).agg({'rank': ['count', 'mean']}).set_axis(['displayLink', 'rank_count', 'rank_mean'], axis=1, inplace=False)
top_df = pd.merge(top_df, top_df_counts_means).sort_values(['rank_count', 'rank_mean'], ascending=[False, True])
rank_counts = top_df.groupby(['displayLink', 'rank']).agg({'rank': ['count']}).reset_index().set_axis(['displayLink', 'rank', 'count'], axis=1, inplace=False)
num_queries = df['queryTime'].nunique()
fig = go.Figure()
fig.layout.height = 600
fig.layout.yaxis.autorange = 'reversed'
fig.layout.yaxis.zeroline = False
for domain in rank_counts['displayLink'].unique():
rank_counts_subset = rank_counts[rank_counts['displayLink'] == domain]
fig.layout.title = 'Google Search Results Rankings<br>keyword(s): ' + str(df['queryTime'].nunique()) + ' Football (Soccer) Queries'
fig.layout.hovermode = False
fig.layout.yaxis.autorange = 'reversed'
fig.layout.yaxis.zeroline = False
fig.layout.yaxis.tickvals = list(range(1, 14))
fig.layout.yaxis.ticktext = list(range(1, 11)) + ['Total<br>appearances', 'Coverage', 'Avg. Pos.']
fig.layout.height = height
fig.layout.width = width
fig.layout.yaxis.title = 'SERP Rank (number of appearances)'
fig.layout.showlegend = False
fig.layout.paper_bgcolor = '#eeeeee'
fig.layout.plot_bgcolor = '#eeeeee'
plot_serps(serp_clubs[serp_clubs['hl'] == 'es'], opacity=0.1) | code |
17134452/cell_14 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
import advertools as adv
import pandas as pd
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key
clubs = pd.read_csv('../input/clubs.csv')
serp_clubs = pd.read_csv('../input/serp_clubs.csv')
serp_clubs['totalResults'] = serp_clubs['totalResults'].astype('int')
serp_clubs['queryTime'] = pd.to_datetime(serp_clubs['queryTime'])
adv.word_frequency(serp_clubs['title'], rm_words=adv.stopwords['english'].union(['-', '|', ' ', ''])).head(10) | code |
17134452/cell_22 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
from plotly.tools import make_subplots
import advertools as adv
import pandas as pd
import plotly.graph_objs as go
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key
clubs = pd.read_csv('../input/clubs.csv')
top_countries = clubs.groupby('Country').agg({'Total': 'sum'}).sort_values('Total', ascending=False).reset_index().head(10)
top_countries
clubs.groupby(['Country']).agg({'Club': 'count', 'Total': 'sum'}).sort_values('Club', ascending=False).reset_index().head(9).set_axis(['country', 'num_clubs', 'total_wins'], axis=1, inplace=False).assign(wins_per_club=lambda df: df['total_wins'].div(df['num_clubs'])).style.background_gradient(high=0.2)
lang_football = {'en': 'football', 'fr': 'football', 'de': 'fußball', 'es': 'fútbol', 'it': 'calcio', 'pt-BR': 'futebol', 'nl': 'voetbal'}
lang_football
serp_clubs = pd.read_csv('../input/serp_clubs.csv')
serp_clubs['totalResults'] = serp_clubs['totalResults'].astype('int')
serp_clubs['queryTime'] = pd.to_datetime(serp_clubs['queryTime'])
club_country = {club.lower(): country.lower() for club, country in zip(clubs['Club'], clubs['Country'])}
football_multi = '|'.join([' ' + football for football in lang_football.values()])
serp_clubs['country'] = [club_country[club].title()
for club in serp_clubs['searchTerms'].str.replace(football_multi, '')]
serp_clubs[['searchTerms', 'country']].sample(10)
serp_clubs.drop_duplicates(['searchTerms']).groupby('searchTerms', as_index=False).agg({'totalResults': 'sum'}).sort_values('totalResults', ascending=False).reset_index(drop=True).head(15).style.format({'totalResults': '{:,}'})
hl_domain_appearances = serp_clubs.groupby(['hl', 'displayLink']).agg({'rank': 'count'}).reset_index().sort_values(['hl', 'rank'], ascending=False).rename(columns={'rank': 'search_appearances'})
hl_domain_appearances.groupby(['hl']).head(5)
fig = make_subplots(1, 7, print_grid=False, shared_yaxes=True)
for i, lang in enumerate(serp_clubs['hl'].unique()[:7]):
df = serp_clubs[serp_clubs['hl']==lang]
fig.append_trace(go.Bar(y=df['displayLink'].value_counts().values[:8],
x=df['displayLink'].value_counts().index.str.replace('www.', '')[:8],
name=lang,
orientation='v'), row=1, col=i+1)
fig.layout.margin = {'b': 150, 'r': 30}
fig.layout.legend.orientation = 'h'
fig.layout.legend.y = -0.5
fig.layout.legend.x = 0.15
fig.layout.title = 'Top Domains by Language of Search'
fig.layout.yaxis.title = 'Number of Appearances on SERPs'
fig.layout.plot_bgcolor = '#eeeeee'
fig.layout.paper_bgcolor = '#eeeeee'
iplot(fig)
fig = make_subplots(1, 7, shared_yaxes=True, print_grid=False)
for i, country in enumerate(serp_clubs['country'].unique()[:7]):
if country in top_countries['Country'][:7].values:
df = serp_clubs[serp_clubs['country'] == country]
fig.append_trace(go.Bar(y=df['displayLink'].value_counts().values[:8], x=df['displayLink'].value_counts().index.str.replace('www.', '')[:8], name=country, orientation='v'), row=1, col=i + 1)
fig.layout.margin = {'b': 150, 'r': 0}
fig.layout.legend.orientation = 'h'
fig.layout.legend.y = -0.5
fig.layout.legend.x = 0.15
fig.layout.title = 'Top Domains by Country of Club'
fig.layout.yaxis.title = 'Number of Appearances on SERPs'
fig.layout.plot_bgcolor = '#eeeeee'
fig.layout.paper_bgcolor = '#eeeeee'
iplot(fig) | code |
17134452/cell_12 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
import advertools as adv
import pandas as pd
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key
clubs = pd.read_csv('../input/clubs.csv')
top_countries = clubs.groupby('Country').agg({'Total': 'sum'}).sort_values('Total', ascending=False).reset_index().head(10)
top_countries
clubs.groupby(['Country']).agg({'Club': 'count', 'Total': 'sum'}).sort_values('Club', ascending=False).reset_index().head(9).set_axis(['country', 'num_clubs', 'total_wins'], axis=1, inplace=False).assign(wins_per_club=lambda df: df['total_wins'].div(df['num_clubs'])).style.background_gradient(high=0.2)
lang_football = {'en': 'football', 'fr': 'football', 'de': 'fußball', 'es': 'fútbol', 'it': 'calcio', 'pt-BR': 'futebol', 'nl': 'voetbal'}
lang_football
serp_clubs = pd.read_csv('../input/serp_clubs.csv')
serp_clubs['totalResults'] = serp_clubs['totalResults'].astype('int')
serp_clubs['queryTime'] = pd.to_datetime(serp_clubs['queryTime'])
club_country = {club.lower(): country.lower() for club, country in zip(clubs['Club'], clubs['Country'])}
football_multi = '|'.join([' ' + football for football in lang_football.values()])
serp_clubs['country'] = [club_country[club].title() for club in serp_clubs['searchTerms'].str.replace(football_multi, '')]
serp_clubs[['searchTerms', 'country']].sample(10) | code |
17134452/cell_5 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from plotly.offline import iplot, init_notebook_mode
import advertools as adv
import pandas as pd
import advertools as adv
import pandas as pd
pd.options.display.max_columns = None
from plotly.tools import make_subplots
import plotly.graph_objs as go
from plotly.offline import iplot, init_notebook_mode
init_notebook_mode()
adv.__version__
column_key = pd.read_csv('../input/column_key.csv')
column_key
clubs = pd.read_csv('../input/clubs.csv')
top_countries = clubs.groupby('Country').agg({'Total': 'sum'}).sort_values('Total', ascending=False).reset_index().head(10)
top_countries | code |
128044650/cell_30 | [
"text_plain_output_2.png",
"text_plain_output_1.png",
"image_output_1.png"
] | !pip install opentsne
from openTSNE import TSNE
if str_data_inf != 'Medulloblastoma Intergrated GSE124814': # Fail in that case by unclear reason
from openTSNE import TSNE
reducer = TSNE(
perplexity=30,
metric="euclidean",
n_jobs=8,
random_state=42,
verbose=True,
)
r = reducer.fit(X)
for (i,j) in [(0,1)]:#,(0,2),(1,2),(2,3),(2,4),(3,4)]:
fig = plt.figure(figsize = (20,12) ); c = 0
ax = sns.scatterplot(x = r[:,i],y=r[:,j], hue = v4color, palette = palette1 , marker = marker1 , alpha = alpha1 )
if 1:
plt.setp(ax.get_legend().get_texts(), fontsize=12) # for legend text
plt.setp(ax.get_legend().get_title(), fontsize=12) # for legend title
plt.title(' openTSNE '+ str_data_inf + ' n_samples='+str(len(r)), fontsize = 20)
plt.show() | code |
128044650/cell_44 | [
"application_vnd.jupyter.stderr_output_2.png",
"text_plain_output_4.png",
"text_plain_output_6.png",
"text_plain_output_3.png",
"application_vnd.jupyter.stderr_output_5.png",
"text_plain_output_7.png",
"text_plain_output_1.png",
"image_output_2.png",
"image_output_1.png"
] | import os
import time
import time
import time
t0start = time.time()
import numpy as np
import pandas as pd
import os
print('%.1f seconds passed total ' % (time.time() - t0start)) | code |
128044650/cell_20 | [
"application_vnd.jupyter.stderr_output_2.png",
"text_plain_output_3.png",
"text_plain_output_1.png",
"image_output_1.png"
] | !pip install trimap
import trimap
reducer = trimap.TRIMAP()
r = reducer.fit_transform(X)
for (i,j) in [(0,1)]:#,(0,2),(1,2),(2,3),(2,4),(3,4)]:
fig = plt.figure(figsize = (20,12) ); c = 0
ax = sns.scatterplot(x = r[:,i],y=r[:,j], hue = v4color, palette = palette1, marker = marker1 , alpha = alpha1 )
if 1:
plt.setp(ax.get_legend().get_texts(), fontsize=20) # for legend text
plt.setp(ax.get_legend().get_title(), fontsize=20) # for legend title
plt.title(' trimap ' + str_data_inf + ' n_samples='+str(len(r)), fontsize = 20)
plt.show() | code |
128044650/cell_40 | [
"text_plain_output_3.png",
"text_plain_output_2.png",
"application_vnd.jupyter.stderr_output_1.png",
"image_output_1.png"
] | from sklearn.manifold import Isomap
fig = plt.figure(figsize=(25, 8))
plt.suptitle(str_data_inf + ' n_samples=' + str(len(X)), fontsize=20)
c = 0
for n_neighbors in [5, 10]:
reducer = Isomap(n_components=2, n_neighbors=n_neighbors)
t0 = time.time()
try:
if isinstance(X, pd.DataFrame):
r = reducer.fit_transform(X.iloc[:N, :])
else:
r = reducer.fit_transform(X[:N, :])
except Exception as e:
print(e)
print('Got Exception', n_neighbors)
continue
t1 = time.time()
print('Isomap n_neighbors %s: %.2g sec' % (n_neighbors, t1 - t0))
c += 1
fig.add_subplot(1, 2, c)
if isinstance(v4color, pd.Series):
sns.scatterplot(x=r[:, 0], y=r[:, 1], hue=v4color.iloc[:N], palette=palette1, marker=marker1, alpha=alpha1)
else:
sns.scatterplot(x=r[:, 0], y=r[:, 1], hue=v4color[:N], palette=palette1, marker=marker1, alpha=alpha1)
plt.title('Isomap n_neighbors' + str(n_neighbors), fontsize=20)
if c == 1:
plt.setp(ax.get_legend().get_texts(), fontsize=12)
plt.setp(ax.get_legend().get_title(), fontsize=12)
plt.legend(bbox_to_anchor=(1.15, 1.0), loc='upper right')
else:
plt.legend('')
plt.show() | code |
128044650/cell_26 | [
"application_vnd.jupyter.stderr_output_2.png",
"text_plain_output_1.png"
] | import umap
reducer = umap.UMAP(densmap=True, random_state=42)
r2 = reducer.fit_transform(X)
print(X.shape)
for i, j in [(0, 1)]:
plt.figure(figsize=(20, 10))
ax = sns.scatterplot(x=r2[:, i], y=r2[:, j], hue=v4color, palette=palette1, marker=marker1, alpha=alpha1)
if 1:
plt.setp(ax.get_legend().get_texts(), fontsize=20)
plt.setp(ax.get_legend().get_title(), fontsize=20)
plt.title('UMAP densmap On ' + str_data_inf + ' n_samples=' + str(len(X)), fontsize=20)
plt.xlabel('UMAP' + str(i + 1), fontsize=12)
plt.ylabel('UMAP' + str(j + 1), fontsize=12)
plt.show() | code |
128044650/cell_2 | [
"text_plain_output_2.png",
"text_plain_output_1.png",
"image_output_1.png"
] | import os
import time
import time
t0start = time.time()
import numpy as np
import pandas as pd
import os
for dirname, _, filenames in os.walk('/kaggle/input'):
for filename in filenames:
if 'png' not in filename:
print(os.path.join(dirname, filename)) | code |
128044650/cell_18 | [
"text_plain_output_1.png",
"image_output_3.png",
"image_output_2.png",
"image_output_1.png"
] | !pip install pacmap
import pacmap
reducer = pacmap.PaCMAP()
r = reducer.fit_transform(X)
for (i,j) in [(0,1)]:#,(0,2),(1,2),(2,3),(2,4),(3,4)]:
fig = plt.figure(figsize = (20,12) ); c = 0
ax = sns.scatterplot(x = r[:,i],y=r[:,j], hue = v4color, palette = palette1, marker = marker1 , alpha = alpha1 )
if 1:
plt.setp(ax.get_legend().get_texts(), fontsize=20) # for legend text
plt.setp(ax.get_legend().get_title(), fontsize=20) # for legend title
plt.title(' pacmap ' + str_data_inf + ' n_samples='+str(len(r)), fontsize = 20)
plt.show() | code |
128044650/cell_32 | [
"text_plain_output_1.png",
"image_output_1.png"
] | import time
import umap
from sklearn import manifold
from sklearn.decomposition import PCA
from sklearn.decomposition import FactorAnalysis
from sklearn.decomposition import NMF
from sklearn.decomposition import FastICA
from sklearn.decomposition import FactorAnalysis
from sklearn.decomposition import LatentDirichletAllocation
from sklearn.ensemble import RandomTreesEmbedding
from sklearn.random_projection import SparseRandomProjection
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis
from sklearn.pipeline import make_pipeline
from sklearn.decomposition import TruncatedSVD
from collections import OrderedDict
from functools import partial
from matplotlib.ticker import NullFormatter
n_neighbors = 10
n_components = 2
LLE = partial(manifold.LocallyLinearEmbedding, n_neighbors, n_components, eigen_solver='auto')
methods = OrderedDict()
methods['PCA'] = PCA()
methods['umap'] = umap.UMAP(n_components=n_components)
methods['t-SNE'] = manifold.TSNE(n_components=n_components, init='pca', random_state=0)
methods['ICA'] = FastICA(n_components=n_components, random_state=0)
methods['FA'] = FactorAnalysis(n_components=n_components, random_state=0)
methods['MDS'] = manifold.MDS(n_components, max_iter=100, n_init=1)
methods['SE'] = manifold.SpectralEmbedding(n_components=n_components, n_neighbors=n_neighbors)
methods['NMF'] = NMF(n_components=n_components, init='random', random_state=0)
methods['RandProj'] = SparseRandomProjection(n_components=n_components, random_state=42)
rand_trees_embed = make_pipeline(RandomTreesEmbedding(n_estimators=200, random_state=0, max_depth=5), TruncatedSVD(n_components=n_components))
methods['RandTrees'] = rand_trees_embed
methods['LatDirAll'] = LatentDirichletAllocation(n_components=n_components, random_state=0)
list_fast_methods = ['FA', 'RandProj', 'RandTrees']
list_slow_methods = ['t-SNE', 'LLE', 'Modified LLE', 'Isomap', 'MDS', 'SE', 'LatDirAll', 'LTSA', 'Hessian LLE']
fig = plt.figure(figsize=(25, 16))
plt.suptitle(str_data_inf + ' n_samples=' + str(len(X)), fontsize=20)
c = 0
for i, (label, method) in enumerate(methods.items()):
if label not in list_fast_methods:
continue
t0 = time.time()
try:
r = method.fit_transform(X)
except:
print('Got Exception', label)
continue
t1 = time.time()
print('%s: %.2g sec' % (label, t1 - t0))
c += 1
fig.add_subplot(2, 4, c)
sns.scatterplot(x=r[:, 0], y=r[:, 1], hue=v4color, palette=palette1, marker=marker1, alpha=alpha1)
plt.title(label, fontsize=20)
if c == 1:
plt.setp(ax.get_legend().get_texts(), fontsize=12)
plt.setp(ax.get_legend().get_title(), fontsize=12)
plt.legend(bbox_to_anchor=(1.15, 1.0), loc='upper right')
else:
plt.legend('')
plt.show() | code |
128044650/cell_28 | [
"text_plain_output_2.png",
"application_vnd.jupyter.stderr_output_1.png",
"image_output_1.png"
] | fig = plt.figure(figsize=(20, 16))
c = 0
cc = 0
plt.suptitle('UMAP densmap ' + str_data_inf + ' n_samples=' + str(len(X)), fontsize=20)
for min_dist in [0.1, 0.9]:
for n_neighbors in [5, 15, 100]:
c += 1
fig.add_subplot(2, 3, c)
str_inf = 'n_neighbors=' + str(n_neighbors) + ' min_dist=' + str(min_dist)
reducer = umap.UMAP(densmap=True, n_neighbors=n_neighbors, min_dist=min_dist, n_components=2)
r2 = reducer.fit_transform(X)
i, j = (0, 1)
ax = sns.scatterplot(x=r2[:, i], y=r2[:, j], hue=v4color, palette=palette1, marker=marker1, alpha=alpha1)
if 1:
plt.setp(ax.get_legend().get_texts(), fontsize=12)
plt.setp(ax.get_legend().get_title(), fontsize=12)
plt.title(str_inf, fontsize=20)
plt.xlabel('UMAP' + str(i + 1), fontsize=20)
plt.ylabel('UMAP' + str(j + 1), fontsize=20)
if c == 1:
plt.setp(ax.get_legend().get_texts(), fontsize=10)
plt.setp(ax.get_legend().get_title(), fontsize=10)
plt.legend(bbox_to_anchor=(1.15, 1.0), loc='upper right')
else:
plt.legend('')
plt.show() | code |
128044650/cell_8 | [
"text_plain_output_1.png"
] | from sklearn.decomposition import PCA
pca = PCA(n_components=6)
r = pca.fit_transform(X)
print(np.sum(pca.explained_variance_ratio_)) | code |
128044650/cell_16 | [
"image_output_5.png",
"image_output_4.png",
"image_output_6.png",
"text_plain_output_1.png",
"image_output_3.png",
"image_output_2.png",
"image_output_1.png"
] | !pip install ncvis
import ncvis
reducer = ncvis.NCVis()
r = reducer.fit_transform(X)
for (i,j) in [(0,1)]:#,(0,2),(1,2),(2,3),(2,4),(3,4)]:
fig = plt.figure(figsize = (20,12) ); c = 0
ax = sns.scatterplot(x = r[:,i],y=r[:,j], hue = v4color, palette = palette1, marker = marker1 , alpha = alpha1 )
if 1:
plt.setp(ax.get_legend().get_texts(), fontsize=20) # for legend text
plt.setp(ax.get_legend().get_title(), fontsize=20) # for legend title
plt.title(' ncvis ' + str_data_inf + ' n_samples='+str(len(r)) , fontsize = 20)
plt.show() | code |
128044650/cell_38 | [
"text_plain_output_2.png",
"text_plain_output_1.png",
"image_output_1.png"
] | from sklearn.manifold import LocallyLinearEmbedding
for n_neighbors in [5, 10]:
print('n_neighbors', n_neighbors)
fig = plt.figure(figsize=(25, 16))
plt.suptitle(str_data_inf + ' n_samples=' + str(len(X)), fontsize=20)
c = 0
for method in ['standard', 'hessian', 'modified', 'ltsa']:
if method == 'hessian':
reducer = LocallyLinearEmbedding(n_components=2, n_neighbors=6, method=method)
else:
reducer = LocallyLinearEmbedding(n_components=2, n_neighbors=n_neighbors, method=method)
t0 = time.time()
try:
if isinstance(X, pd.DataFrame):
r = reducer.fit_transform(X.iloc[:N, :])
else:
r = reducer.fit_transform(X[:N, :])
except Exception as e:
print(e)
print('Got Exception', method)
continue
t1 = time.time()
print('LLE %s: %.2g sec' % (method, t1 - t0))
c += 1
fig.add_subplot(2, 2, c)
if isinstance(v4color, pd.Series):
sns.scatterplot(x=r[:, 0], y=r[:, 1], hue=v4color.iloc[:N], palette=palette1, marker=marker1, alpha=alpha1)
else:
sns.scatterplot(x=r[:, 0], y=r[:, 1], hue=v4color[:N], palette=palette1, marker=marker1, alpha=alpha1)
plt.title('LLE ' + str(method), fontsize=20)
plt.legend('')
plt.show() | code |
128044650/cell_24 | [
"text_plain_output_2.png",
"text_plain_output_1.png",
"image_output_1.png"
] | fig = plt.figure(figsize=(20, 16))
c = 0
cc = 0
plt.suptitle('UMAP ' + str_data_inf + ' n_samples=' + str(len(X)), fontsize=20)
for min_dist in [0.1, 0.9]:
for n_neighbors in [5, 15, 100]:
c += 1
fig.add_subplot(2, 3, c)
str_inf = 'n_neighbors=' + str(n_neighbors) + ' min_dist=' + str(min_dist)
reducer = umap.UMAP(n_neighbors=n_neighbors, min_dist=min_dist, n_components=2)
r2 = reducer.fit_transform(X)
i, j = (0, 1)
ax = sns.scatterplot(x=r2[:, i], y=r2[:, j], hue=v4color, palette=palette1, marker=marker1, alpha=alpha1)
if 1:
plt.setp(ax.get_legend().get_texts(), fontsize=12)
plt.setp(ax.get_legend().get_title(), fontsize=12)
plt.title(str_inf, fontsize=20)
plt.xlabel('UMAP' + str(i + 1), fontsize=20)
plt.ylabel('UMAP' + str(j + 1), fontsize=20)
if c == 1:
plt.setp(ax.get_legend().get_texts(), fontsize=10)
plt.setp(ax.get_legend().get_title(), fontsize=10)
plt.legend(bbox_to_anchor=(1.15, 1.0), loc='upper right')
else:
plt.legend('')
plt.show() | code |
128044650/cell_14 | [
"text_plain_output_1.png"
] | import matplotlib.pyplot as plt
import seaborn as sns
import umap
r = umap.UMAP().fit_transform(X)
print(r.shape)
for i, j in [(0, 1)]:
sns.scatterplot(x=r[:, i], y=r[:, j], hue=v4color, palette='rainbow')
plt.xlabel('UMAP' + str(i + 1), fontsize=20)
plt.ylabel('UMAP' + str(j + 1), fontsize=20)
plt.show() | code |
128044650/cell_22 | [
"text_plain_output_2.png",
"text_plain_output_1.png",
"image_output_1.png"
] | !pip install MulticoreTSNE
from MulticoreTSNE import MulticoreTSNE as TSNE
reducer = TSNE(n_jobs=4)
r = reducer.fit_transform(X)
for (i,j) in [(0,1)]:#,(0,2),(1,2),(2,3),(2,4),(3,4)]:
fig = plt.figure(figsize = (20,12) ); c = 0
ax = sns.scatterplot(x = r[:,i],y=r[:,j], hue = v4color, palette = palette1 , marker = marker1 , alpha = alpha1 )
if 1:
plt.setp(ax.get_legend().get_texts(), fontsize=20) # for legend text
plt.setp(ax.get_legend().get_title(), fontsize=20) # for legend title
plt.title(' MulticoreTSNE ' + str_data_inf + ' n_samples='+str(len(r)), fontsize = 20)
plt.show() | code |
128044650/cell_10 | [
"text_plain_output_1.png"
] | n_x_subplots = 2
c = 0
cc = 0
for i, j in [(0, 1), (0, 2), (0, 3), (1, 2), (1, 3), (2, 3), (0, 4), (1, 4), (2, 4), (3, 4), (0, 5)]:
cc += 1
if c % n_x_subplots == 0:
if c > 0:
plt.show()
fig = plt.figure(figsize=(20, 5))
c = 0
plt.suptitle(str_data_inf + ' n_samples=' + str(len(r)), fontsize=20)
c += 1
fig.add_subplot(1, n_x_subplots, c)
ax = sns.scatterplot(x=r[:, i], y=r[:, j], hue=v4color, palette=palette1, marker=marker1, alpha=alpha1)
if c == 1:
plt.setp(ax.get_legend().get_texts(), fontsize=12)
plt.setp(ax.get_legend().get_title(), fontsize=12)
plt.legend(bbox_to_anchor=(1.15, 1.0), loc='upper right')
else:
plt.legend('')
plt.title('PCA ' + str(i) + ' ' + str(j), fontsize=20)
plt.show() | code |
128044650/cell_12 | [
"text_plain_output_2.png",
"text_plain_output_1.png"
] | from sklearn.decomposition import FastICA
reducer = FastICA(n_components=5, random_state=0, whiten='unit-variance')
r = reducer.fit_transform(X)
n_x_subplots = 2
c = 0
cc = 0
for i, j in [(0, 1), (0, 2), (1, 2), (2, 3), (2, 4), (3, 4)]:
if c % n_x_subplots == 0:
if c > 0:
plt.show()
fig = plt.figure(figsize=(20, 5))
c = 0
plt.suptitle(str_data_inf + ' n_samples=' + str(len(r)), fontsize=20)
c += 1
fig.add_subplot(1, n_x_subplots, c)
ax = sns.scatterplot(x=r[:, i], y=r[:, j], hue=v4color, palette=palette1, marker=marker1, alpha=alpha1)
if 1:
plt.setp(ax.get_legend().get_texts(), fontsize=12)
plt.setp(ax.get_legend().get_title(), fontsize=12)
plt.title('ICA ' + str(i) + ' ' + str(j), fontsize=20)
if c == 1:
plt.setp(ax.get_legend().get_texts(), fontsize=12)
plt.setp(ax.get_legend().get_title(), fontsize=12)
plt.legend(bbox_to_anchor=(1.15, 1.0), loc='upper right')
else:
plt.legend('')
plt.show() | code |
128044650/cell_5 | [
"text_plain_output_2.png",
"text_plain_output_1.png",
"image_output_1.png"
] | features = 'esm2_t33_650M'
if features == 'T5':
fn = '/kaggle/input/t5embeds/train_embeds.npy'
fn4submit = '/kaggle/input/t5embeds/test_embeds.npy'
str_data_inf = 'CAFA5 T5 embeddings '
elif features == 'esm2_t33_650M':
fn = '/kaggle/input/23468234/train_embeds_esm2_t33_650M_UR50D.npy'
fn4submit = '/kaggle/input/23468234/test_embeds_esm2_t33_650M_UR50D.npy'
str_data_inf = 'CAFA5 ' + features
elif features == 'esm2_t30_150M':
fn = '/kaggle/input/8947923/train_embeds_esm2_t30_150M_UR50D.npy'
fn4submit = '/kaggle/input/8947923/test_embeds_esm2_t30_150M_UR50D.npy'
str_data_inf = 'CAFA5 ' + features
elif features == 'esm2_t12_35M':
fn = '/kaggle/input/3023750/train_embeds_esm2_t12_35M_UR50D.npy'
fn4submit = '/kaggle/input/3023750/test_embeds_esm2_t12_35M_UR50D.npy'
str_data_inf = 'CAFA5 ' + features
elif features == 'esm2_t6_8M':
fn = '/kaggle/input/315701375/train_embeds_esm2_t6_8M_UR50D.npy'
fn4submit = '/kaggle/input/315701375/test_embeds_esm2_t6_8M_UR50D.npy'
str_data_inf = 'CAFA5 ' + features
elif features == 'SGTv12':
fn = '/kaggle/input/cafa5-sgt-protein-embeddings/run12_fit_on_15000x2/sgt_embendings_train_15000_15000.csv'
fn4submit = '/kaggle/input/cafa5-sgt-protein-embeddings/run12_fit_on_15000x2/sgt_embendings_test_15000_15000.csv'
str_data_inf = 'CAFA5 SGT embeddings v12 '
elif features == 'Leven5000':
fn = '/kaggle/input/cafa5-levenshtein-distance-features-big/run9_achnor_set5000/df_Levenshtein_distance_5000_features_train.csv'
fn4submit = '/kaggle/input/cafa5-levenshtein-distance-features-big/run9_achnor_set5000/df_Levenshtein_distance_5000_features_test.csv'
str_data_inf = 'CAFA5 Levenshtein distance features anchor 5000 '
elif features == 'Leven3000':
fn = '/kaggle/input/cafa5-levenshtein-distance-features-big/run8_anchor_set_3000/df_Levenshtein_distance_3000_features_train.csv'
fn4submit = '/kaggle/input/cafa5-levenshtein-distance-features-big/run8_anchor_set_3000/df_Levenshtein_distance_3000_features_test.csv'
str_data_inf = 'CAFA5 Levenshtein distance features anchor 3000 '
elif features == 'Leven1000':
fn = '/kaggle/input/cafa5-features-etc/df_Levenshtein_distance_1000_features_train.csv'
fn4submit = '/kaggle/input/cafa5-features-etc/df_Levenshtein_distance_1000_features_test.csv'
str_data_inf = 'CAFA5 Levenshtein distance features anchor 1000 '
elif features == 'Leven100':
fn = '/kaggle/input/cafa5-features-etc/df_Levenshtein_distance_100_features_train.csv'
fn4submit = '/kaggle/input/cafa5-features-etc/df_Levenshtein_distance_100_features_test.csv'
str_data_inf = 'CAFA5 Levenshtein distance features anchor 100 '
def load_features(fn):
print(fn)
if '.csv' in fn:
df = pd.read_csv(fn, index_col=0)
X = df.values
elif '.npy' in fn:
X = np.load(fn)
print(X.shape)
return X
X = load_features(fn)
str_data_inf += ' train'
print(X.shape)
X[:10, :10] | code |
128044650/cell_36 | [
"text_plain_output_2.png",
"text_plain_output_1.png",
"image_output_1.png"
] | import umap
from sklearn import manifold
from sklearn.decomposition import PCA
from sklearn.decomposition import FactorAnalysis
from sklearn.decomposition import NMF
from sklearn.decomposition import FastICA
from sklearn.decomposition import FactorAnalysis
from sklearn.decomposition import LatentDirichletAllocation
from sklearn.ensemble import RandomTreesEmbedding
from sklearn.random_projection import SparseRandomProjection
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis
from sklearn.pipeline import make_pipeline
from sklearn.decomposition import TruncatedSVD
from collections import OrderedDict
from functools import partial
from matplotlib.ticker import NullFormatter
n_neighbors = 10
n_components = 2
LLE = partial(manifold.LocallyLinearEmbedding, n_neighbors, n_components, eigen_solver='auto')
methods = OrderedDict()
methods['PCA'] = PCA()
methods['umap'] = umap.UMAP(n_components=n_components)
methods['t-SNE'] = manifold.TSNE(n_components=n_components, init='pca', random_state=0)
methods['ICA'] = FastICA(n_components=n_components, random_state=0)
methods['FA'] = FactorAnalysis(n_components=n_components, random_state=0)
methods['MDS'] = manifold.MDS(n_components, max_iter=100, n_init=1)
methods['SE'] = manifold.SpectralEmbedding(n_components=n_components, n_neighbors=n_neighbors)
methods['NMF'] = NMF(n_components=n_components, init='random', random_state=0)
methods['RandProj'] = SparseRandomProjection(n_components=n_components, random_state=42)
rand_trees_embed = make_pipeline(RandomTreesEmbedding(n_estimators=200, random_state=0, max_depth=5), TruncatedSVD(n_components=n_components))
methods['RandTrees'] = rand_trees_embed
methods['LatDirAll'] = LatentDirichletAllocation(n_components=n_components, random_state=0)
list_fast_methods = ['PCA', 'umap', 'FA', 'NMF', 'RandProj', 'RandTrees']
list_slow_methods = ['t-SNE', 'LLE', 'Modified LLE', 'Isomap', 'MDS', 'SE', 'LatDirAll', 'LTSA', 'Hessian LLE']
fig = plt.figure(figsize=(25, 16))
plt.suptitle(str_data_inf + ' n_samples=' + str(len(X)), fontsize=20)
c = 0
for i, (label, method) in enumerate(methods.items()):
if label not in list_slow_methods:
continue
t0 = time.time()
try:
if isinstance(X, pd.DataFrame):
r = method.fit_transform(X.iloc[:N, :])
else:
r = method.fit_transform(X[:N, :])
except:
print('Got Exception', label)
continue
t1 = time.time()
print('%s: %.2g sec' % (label, t1 - t0))
c += 1
fig.add_subplot(2, 3, c)
if isinstance(v4color, pd.Series):
sns.scatterplot(x=r[:, 0], y=r[:, 1], hue=v4color.iloc[:N], palette=palette1, marker=marker1, alpha=alpha1)
else:
sns.scatterplot(x=r[:, 0], y=r[:, 1], hue=v4color[:N], palette=palette1, marker=marker1, alpha=alpha1)
plt.title(label, fontsize=20)
if c == 1:
plt.setp(ax.get_legend().get_texts(), fontsize=12)
plt.setp(ax.get_legend().get_title(), fontsize=12)
plt.legend(bbox_to_anchor=(1.15, 1.0), loc='upper right')
else:
plt.legend('')
plt.show() | code |
105199203/cell_4 | [
"text_plain_output_1.png"
] | n1 = int(input('Enter your number 1'))
n2 = int(input('Enter your number 2'))
n3 = int(input('Enter your number 3'))
max = n1
if max < n2:
max = n2
if max < n3:
max = n3
if n1 > n2 and n1 > n3:
print(n1, 'Is the maximun value')
elif n2 > n1 and n2 > n3:
print(n2, 'Is the maximun value')
elif n3 > n1 and n3 > n2:
print(n3, 'Is the maximun value') | code |
105199203/cell_3 | [
"text_plain_output_1.png"
] | n1 = int(input('Enter your number 1'))
n2 = int(input('Enter your number 2'))
n3 = int(input('Enter your number 3'))
max = n1
if max < n2:
max = n2
if max < n3:
max = n3
print(max, 'Is the maximun number') | code |
122245369/cell_1 | [
"text_plain_output_1.png"
] | import os
import numpy as np
import pandas as pd
import os
for dirname, _, filenames in os.walk('/kaggle/input'):
for filename in filenames:
print(os.path.join(dirname, filename)) | code |
122245369/cell_7 | [
"text_plain_output_2.png",
"text_plain_output_1.png"
] | import torchvision
import torchvision.transforms as transforms
train_transformer = transforms.Compose([transforms.RandomCrop(32, padding=5), transforms.RandomHorizontalFlip(p=0.5), transforms.RandomRotation(degrees=20), transforms.ToTensor()])
test_transformer = transforms.Compose([transforms.ToTensor()])
train_dataset = torchvision.datasets.CIFAR10(root='.', train=True, transform=train_transformer, download=True)
test_dataset = torchvision.datasets.CIFAR10(root='.', train=False, transform=test_transformer, download=True) | code |
122246401/cell_13 | [
"text_plain_output_1.png"
] | from keras.preprocessing.image import ImageDataGenerator,load_img
load_img('../input/rice-image-dataset/Rice_Image_Dataset/Basmati/basmati (10009).jpg') | code |
122246401/cell_9 | [
"text_plain_output_1.png"
] | from keras.preprocessing.image import ImageDataGenerator,load_img
from tensorflow import keras
from tensorflow.keras.layers import Dense,Flatten
from tensorflow.keras.models import Sequential
train_gen = ImageDataGenerator(rescale=1 / 255.0, validation_split=0.3)
train_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='training')
test_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='validation')
train_data.class_indices
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Flatten
model = Sequential()
model.add(Flatten(input_shape=(64, 64, 3)))
model.add(Dense(40, activation='sigmoid'))
model.add(Dense(5, activation='relu'))
model.add(Dense(5, activation='softmax'))
model.summary()
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics='accuracy')
model.fit(train_data, validation_data=test_data, epochs=2)
model = keras.models.load_model('../input/rice-classification/cnn_model.h5')
model.summary() | code |
122246401/cell_4 | [
"text_plain_output_1.png"
] | from keras.preprocessing.image import ImageDataGenerator,load_img
train_gen = ImageDataGenerator(rescale=1 / 255.0, validation_split=0.3)
train_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='training')
test_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='validation') | code |
122246401/cell_6 | [
"image_output_1.png"
] | from tensorflow.keras.layers import Dense,Flatten
from tensorflow.keras.models import Sequential
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Flatten
model = Sequential()
model.add(Flatten(input_shape=(64, 64, 3)))
model.add(Dense(40, activation='sigmoid'))
model.add(Dense(5, activation='relu'))
model.add(Dense(5, activation='softmax'))
model.summary() | code |
122246401/cell_2 | [
"text_plain_output_1.png"
] | from keras.preprocessing.image import ImageDataGenerator,load_img
load_img('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/Basmati/basmati (6626).jpg', target_size=(180, 180)) | code |
122246401/cell_11 | [
"text_plain_output_1.png"
] | from keras.preprocessing.image import ImageDataGenerator,load_img
from tensorflow import keras
from tensorflow.keras.layers import Dense,Flatten
from tensorflow.keras.models import Sequential
train_gen = ImageDataGenerator(rescale=1 / 255.0, validation_split=0.3)
train_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='training')
test_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='validation')
train_data.class_indices
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Flatten
model = Sequential()
model.add(Flatten(input_shape=(64, 64, 3)))
model.add(Dense(40, activation='sigmoid'))
model.add(Dense(5, activation='relu'))
model.add(Dense(5, activation='softmax'))
model.summary()
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics='accuracy')
model.fit(train_data, validation_data=test_data, epochs=2)
model = keras.models.load_model('../input/rice-classification/cnn_model.h5')
model.summary()
model.evaluate(test_data)
predict1 = model.predict(test_data).argmax(axis=1)
predict1 | code |
122246401/cell_8 | [
"image_output_1.png"
] | from keras.preprocessing.image import ImageDataGenerator,load_img
from tensorflow.keras.layers import Dense,Flatten
from tensorflow.keras.models import Sequential
train_gen = ImageDataGenerator(rescale=1 / 255.0, validation_split=0.3)
train_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='training')
test_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='validation')
train_data.class_indices
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Flatten
model = Sequential()
model.add(Flatten(input_shape=(64, 64, 3)))
model.add(Dense(40, activation='sigmoid'))
model.add(Dense(5, activation='relu'))
model.add(Dense(5, activation='softmax'))
model.summary()
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics='accuracy')
model.fit(train_data, validation_data=test_data, epochs=2) | code |
122246401/cell_15 | [
"text_plain_output_2.png",
"text_plain_output_1.png"
] | from keras.preprocessing.image import ImageDataGenerator,load_img
from tensorflow import keras
from tensorflow import keras,lite
from tensorflow.keras.layers import Dense,Flatten
from tensorflow.keras.models import Sequential
import cv2
import numpy as np
train_gen = ImageDataGenerator(rescale=1 / 255.0, validation_split=0.3)
train_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='training')
test_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='validation')
train_data.class_indices
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Flatten
model = Sequential()
model.add(Flatten(input_shape=(64, 64, 3)))
model.add(Dense(40, activation='sigmoid'))
model.add(Dense(5, activation='relu'))
model.add(Dense(5, activation='softmax'))
model.summary()
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics='accuracy')
model.fit(train_data, validation_data=test_data, epochs=2)
model = keras.models.load_model('../input/rice-classification/cnn_model.h5')
model.summary()
model.evaluate(test_data)
predict1 = model.predict(test_data).argmax(axis=1)
predict1
import cv2
import numpy as np
image = cv2.imread('../input/rice-img-test/jas3.jpg')
image = cv2.resize(image, (64, 64))
image = image / 255
image = image.reshape(-1, 64, 64, 3)
np.round(model.predict(image)).argmax(axis=1)
from tensorflow import keras, lite
converter = lite.TFLiteConverter.from_keras_model(model)
tfmodel = converter.convert()
open('linear.tflite', 'wb').write(tfmodel) | code |
122246401/cell_3 | [
"text_plain_output_2.png",
"text_plain_output_1.png"
] | from keras.preprocessing.image import ImageDataGenerator,load_img
train_gen = ImageDataGenerator(rescale=1 / 255.0, validation_split=0.3)
train_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='training') | code |
122246401/cell_14 | [
"text_plain_output_1.png"
] | from keras.preprocessing.image import ImageDataGenerator,load_img
load_img('../input/rice-image-dataset/Rice_Image_Dataset/Jasmine/Jasmine (10004).jpg') | code |
122246401/cell_10 | [
"image_output_1.png"
] | from keras.preprocessing.image import ImageDataGenerator,load_img
from tensorflow import keras
from tensorflow.keras.layers import Dense,Flatten
from tensorflow.keras.models import Sequential
train_gen = ImageDataGenerator(rescale=1 / 255.0, validation_split=0.3)
train_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='training')
test_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='validation')
train_data.class_indices
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Flatten
model = Sequential()
model.add(Flatten(input_shape=(64, 64, 3)))
model.add(Dense(40, activation='sigmoid'))
model.add(Dense(5, activation='relu'))
model.add(Dense(5, activation='softmax'))
model.summary()
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics='accuracy')
model.fit(train_data, validation_data=test_data, epochs=2)
model = keras.models.load_model('../input/rice-classification/cnn_model.h5')
model.summary()
model.evaluate(test_data) | code |
122246401/cell_12 | [
"text_plain_output_1.png"
] | from keras.preprocessing.image import ImageDataGenerator,load_img
from tensorflow import keras
from tensorflow.keras.layers import Dense,Flatten
from tensorflow.keras.models import Sequential
import cv2
import numpy as np
train_gen = ImageDataGenerator(rescale=1 / 255.0, validation_split=0.3)
train_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='training')
test_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='validation')
train_data.class_indices
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Flatten
model = Sequential()
model.add(Flatten(input_shape=(64, 64, 3)))
model.add(Dense(40, activation='sigmoid'))
model.add(Dense(5, activation='relu'))
model.add(Dense(5, activation='softmax'))
model.summary()
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics='accuracy')
model.fit(train_data, validation_data=test_data, epochs=2)
model = keras.models.load_model('../input/rice-classification/cnn_model.h5')
model.summary()
model.evaluate(test_data)
predict1 = model.predict(test_data).argmax(axis=1)
predict1
import cv2
import numpy as np
image = cv2.imread('../input/rice-img-test/jas3.jpg')
image = cv2.resize(image, (64, 64))
image = image / 255
image = image.reshape(-1, 64, 64, 3)
np.round(model.predict(image)).argmax(axis=1) | code |
122246401/cell_5 | [
"text_plain_output_1.png"
] | from keras.preprocessing.image import ImageDataGenerator,load_img
train_gen = ImageDataGenerator(rescale=1 / 255.0, validation_split=0.3)
train_data = train_gen.flow_from_directory('/kaggle/input/rice-image-dataset/Rice_Image_Dataset/', target_size=(64, 64), batch_size=1, class_mode='categorical', shuffle=False, subset='training')
train_data.class_indices | code |
16116674/cell_63 | [
"text_plain_output_1.png"
] | from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_absolute_error, mean_squared_error, r2_score
import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df.corr()
df_categorical_col = df.select_dtypes(exclude=np.number).columns
df_categorical_col
df_numeric_col = df.select_dtypes(include=np.number).columns
df_numeric_col
model = LinearRegression()
model.fit(train_x, train_y)
print('Predicting train data')
train_predict = model.predict(train_x)
print('Predicting test data')
test_predict = model.predict(test_x)
print(' ')
print('MAE')
print('Train data: ', mean_absolute_error(train_y, train_predict))
print('Test data: ', mean_absolute_error(test_y, test_predict))
print(' ')
print('MSE')
print('Train data: ', mean_squared_error(train_y, train_predict))
print('Test data: ', mean_squared_error(test_y, test_predict))
print(' ')
print('RMSE')
print('Train data: ', np.sqrt(mean_squared_error(train_y, train_predict)))
print('Test data: ', np.sqrt(mean_squared_error(test_y, test_predict)))
print(' ')
print('R^2')
print('Train data: ', r2_score(train_y, train_predict))
print('Test data: ', r2_score(test_y, test_predict)) | code |
16116674/cell_21 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts() | code |
16116674/cell_25 | [
"text_html_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df | code |
16116674/cell_4 | [
"text_plain_output_1.png",
"image_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.head() | code |
16116674/cell_30 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import seaborn as sns
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df.corr()
sns.pairplot(data=df) | code |
16116674/cell_44 | [
"text_plain_output_1.png",
"image_output_1.png"
] | import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df.corr()
df_categorical_col = df.select_dtypes(exclude=np.number).columns
df_categorical_col
df_numeric_col = df.select_dtypes(include=np.number).columns
df_numeric_col | code |
16116674/cell_40 | [
"text_plain_output_1.png",
"image_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import seaborn as sns
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df.corr()
sns.swarmplot(x=df['smoker'], y=df['expenses']) | code |
16116674/cell_39 | [
"text_plain_output_1.png",
"image_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import seaborn as sns
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df.corr()
sns.lmplot(x='bmi', y='expenses', hue='smoker', data=df) | code |
16116674/cell_60 | [
"text_plain_output_1.png"
] | from sklearn.linear_model import LinearRegression
model = LinearRegression()
model.fit(train_x, train_y)
print(model.coef_) | code |
16116674/cell_19 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique() | code |
16116674/cell_1 | [
"text_plain_output_1.png"
] | import os
import numpy as np
import pandas as pd
import os
print(os.listdir('../input')) | code |
16116674/cell_7 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape | code |
16116674/cell_49 | [
"text_plain_output_1.png"
] | import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df.corr()
df_categorical_col = df.select_dtypes(exclude=np.number).columns
df_categorical_col
df_numeric_col = df.select_dtypes(include=np.number).columns
df_numeric_col
df_onehot = pd.get_dummies(df[df_categorical_col])
df_after_encoding = pd.concat([df[df_numeric_col], df_onehot], axis=1)
df_after_encoding | code |
16116674/cell_32 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import seaborn as sns
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df.corr()
sns.regplot(x=df['age'], y=df['expenses']) | code |
16116674/cell_59 | [
"text_html_output_1.png"
] | from sklearn.linear_model import LinearRegression
model = LinearRegression()
model.fit(train_x, train_y)
print(model.intercept_) | code |
16116674/cell_58 | [
"text_html_output_1.png"
] | from sklearn.linear_model import LinearRegression
model = LinearRegression()
model.fit(train_x, train_y) | code |
16116674/cell_8 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.info() | code |
16116674/cell_15 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df | code |
16116674/cell_47 | [
"text_plain_output_1.png",
"image_output_1.png"
] | import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df.corr()
df_categorical_col = df.select_dtypes(exclude=np.number).columns
df_categorical_col
df_numeric_col = df.select_dtypes(include=np.number).columns
df_numeric_col
df_onehot = pd.get_dummies(df[df_categorical_col])
df_onehot | code |
16116674/cell_17 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique() | code |
16116674/cell_35 | [
"text_html_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import seaborn as sns
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df.corr()
sns.scatterplot(x=df['bmi'], y=df['expenses'], hue=df['smoker']) | code |
16116674/cell_43 | [
"text_plain_output_1.png",
"image_output_1.png"
] | import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df.corr()
df_categorical_col = df.select_dtypes(exclude=np.number).columns
df_categorical_col | code |
16116674/cell_14 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum() | code |
16116674/cell_10 | [
"text_html_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum() | code |
16116674/cell_27 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df.corr() | code |
16116674/cell_37 | [
"text_html_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import seaborn as sns
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum()
df = df.drop_duplicates()
df.duplicated().sum()
df.sex.unique()
df.region.unique()
df.smoker.value_counts()
df.smoker.replace({'no': 0, 'yes': 1}, inplace=True)
df.corr()
sns.scatterplot(x=df['age'], y=df['expenses'], hue=df['smoker']) | code |
16116674/cell_12 | [
"text_html_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T
df.shape
df.isna().sum()
df.duplicated().sum() | code |
16116674/cell_5 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
df = pd.read_csv('../input/insurance.csv')
df.describe().T | code |
33121549/cell_13 | [
"text_html_output_1.png"
] | from tensorflow import keras
from tensorflow.keras import layers
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import tensorflow as tf
import tensorflow_docs as tfdocs
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
tf.debugging.set_log_device_placement(False)
dataset = pd.read_csv('/kaggle/input/pubg-finish-placement-prediction/train_V2.csv')[['damageDealt', 'headshotKills', 'killPlace', 'boosts', 'heals', 'winPlacePerc']].dropna()
train_dataset = dataset.sample(frac=0.9, random_state=0)
test_dataset = dataset.drop(train_dataset.index)
train_stats = train_dataset.describe()
train_stats.pop("winPlacePerc")
train_stats = train_stats.transpose()
train_stats
train_labels = train_dataset.pop('winPlacePerc')
test_labels = test_dataset.pop('winPlacePerc')
def norm(x):
return (x - train_stats['mean']) / train_stats['std']
normed_train_data = norm(train_dataset)
normed_test_data = norm(test_dataset)
def build_model():
model = keras.Sequential([layers.Dense(64, activation='relu', input_shape=(len(train_dataset.keys()),)), layers.Dense(64, activation='relu'), layers.Dense(1)])
optimizer = tf.keras.optimizers.Adam(learning_rate=0.001, beta_1=0.9, beta_2=0.999, amsgrad=False)
model.compile(loss='mae', optimizer=optimizer, metrics=['mae', 'mse'])
return model
model = build_model()
early_stop = keras.callbacks.EarlyStopping(monitor='val_loss', patience=10)
EPOCHS = 2
history = model.fit(normed_train_data, train_labels, epochs=EPOCHS, validation_split=0.2, verbose=1, batch_size=300, callbacks=[early_stop, tfdocs.modeling.EpochDots()]) | code |
33121549/cell_1 | [
"text_plain_output_1.png"
] | import os
import numpy as np
import pandas as pd
import os
for dirname, _, filenames in os.walk('/kaggle/input'):
for filename in filenames:
print(os.path.join(dirname, filename)) | code |
33121549/cell_8 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
dataset = pd.read_csv('/kaggle/input/pubg-finish-placement-prediction/train_V2.csv')[['damageDealt', 'headshotKills', 'killPlace', 'boosts', 'heals', 'winPlacePerc']].dropna()
train_dataset = dataset.sample(frac=0.9, random_state=0)
test_dataset = dataset.drop(train_dataset.index)
train_stats = train_dataset.describe()
train_stats.pop('winPlacePerc')
train_stats = train_stats.transpose()
train_stats | code |
33121549/cell_15 | [
"text_plain_output_1.png"
] | """
test_predictions = model.predict(normed_test_data).flatten()
a = plt.axes(aspect='equal')
plt.scatter(test_labels, test_predictions)
plt.xlabel('True Values')
plt.ylabel('Predictions')
lims = [0, 1]
plt.xlim(lims)
plt.ylim(lims)
_ = plt.plot(lims, lims)
""" | code |
33121549/cell_16 | [
"text_plain_output_1.png"
] | """
error = test_predictions - test_labels
plt.hist(error, bins = 25)
plt.xlabel("Prediction Error")
_ = plt.ylabel("Count")
""" | code |
34141750/cell_6 | [
"text_html_output_1.png"
] | import os
import numpy as np
import pandas as pd
import os
for dirname, _, filenames in os.walk('/kaggle/input'):
for filename in filenames:
print(os.path.join(dirname, filename)) | code |
34141750/cell_11 | [
"text_plain_output_1.png"
] | # List /kaggle/input để xem các files và folders dữ liệu được liên kết với Notebook
!ls /kaggle/input | code |
34141750/cell_14 | [
"text_html_output_1.png",
"application_vnd.jupyter.stderr_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
data = pd.read_csv('/kaggle/input/dataisbeautiful/r_dataisbeautiful_posts.csv')
pd.read_csv('/kaggle/temp/temp.csv') | code |
34141750/cell_12 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
data = pd.read_csv('/kaggle/input/dataisbeautiful/r_dataisbeautiful_posts.csv')
data.head(3) | code |
129034049/cell_4 | [
"application_vnd.jupyter.stderr_output_1.png"
] | from bark import SAMPLE_RATE, generate_audio, preload_models
from bark import SAMPLE_RATE, generate_audio, preload_models
from IPython.display import Audio
preload_models() | code |
129034049/cell_2 | [
"text_plain_output_1.png"
] | # install bark as well as pytorch nightly to get blazing fast flash-attention
!pip install git+https://github.com/suno-ai/bark.git && \
pip uninstall -y torch torchvision torchaudio && \
pip install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu118 | code |
128018796/cell_2 | [
"text_plain_output_4.png",
"application_vnd.jupyter.stderr_output_3.png",
"application_vnd.jupyter.stderr_output_5.png",
"text_plain_output_2.png",
"application_vnd.jupyter.stderr_output_1.png"
] | !pip install xmltodict | code |
128018796/cell_8 | [
"application_vnd.jupyter.stderr_output_2.png",
"text_plain_output_1.png"
] | from pathlib import Path
from tqdm import tqdm
from tsfresh import extract_features
import datetime
import numpy as np
import pandas as pd
import re
import xmltodict
import datetime
import json
import re
from os import listdir
from os.path import isfile, join
from pathlib import Path
import catboost
import numpy as np
import optuna
import pandas as pd
import shap
import xmltodict
from boruta import BorutaPy
from dateutil.relativedelta import relativedelta
from sklearn.ensemble import RandomForestRegressor
from tqdm import tqdm
pd.options.mode.chained_assignment = None
DATA_ROOT = Path('/kaggle/input/kaggle-pog-series-s01e04/')
DF_TRAIN = DATA_ROOT / 'train.csv'
DF_TEST = DATA_ROOT / 'test.csv'
RAW_DATA = DATA_ROOT / 'raw_health_export.xml'
OUTPUT_PATH = '/kaggle/working/'
TSFRESH_PATH = '/kaggle/input/datasets-sleep/'
train = pd.read_csv(DF_TRAIN)
test = pd.read_csv(DF_TEST)
class FeatureExtractor:
def __init__(self, data_path):
raw_dict = FeatureExtractor.parse_raw_data(data_path)
self.records = FeatureExtractor.get_records_df(raw_dict)
@staticmethod
def parse_raw_data(data_path):
with open(RAW_DATA) as file:
raw_dict = xmltodict.parse(file.read())
return raw_dict
@staticmethod
def get_records_df(raw_dict):
records = pd.DataFrame(raw_dict['HealthData']['Record'])
records.columns = [col.replace('@', '') for col in records.columns]
records['creationDate'] = pd.to_datetime(records['creationDate']).dt.tz_localize(None)
records['startDate'] = pd.to_datetime(records['startDate']).dt.tz_localize(None)
records['endDate'] = pd.to_datetime(records['endDate']).dt.tz_localize(None)
records['type'] = records['type'].str.replace('HKQuantityTypeIdentifier', '')
return records
@staticmethod
def get_type_ts(records, type):
if type != 'BoostedHeartRate':
records['is_apple'] = records['sourceName'].str.contains('Apple')
type_records = records.query(f"type == '{type}' and is_apple")
type_records = type_records.drop(['type'], axis=1)
type_records['value'] = type_records['value'].astype(float)
else:
hrv = records.query('HeartRateVariabilityMetadataList == HeartRateVariabilityMetadataList')
extract_list = lambda x: x['InstantaneousBeatsPerMinute']
hrv['HeartRateVariabilityMetadataList'] = hrv['HeartRateVariabilityMetadataList'].apply(extract_list)
hrv_exp = hrv.explode('HeartRateVariabilityMetadataList')
get_bpm = lambda x: x['@bpm']
get_time = lambda x: x['@time']
hrv_exp['bpm'] = hrv_exp['HeartRateVariabilityMetadataList'].apply(get_bpm)
hrv_exp['time'] = hrv_exp['HeartRateVariabilityMetadataList'].apply(get_time)
hrv_exp['time_date'] = pd.to_datetime(hrv_exp['time'], format='%I:%M:%S.%f %p')
hrv_exp.loc[hrv_exp['time_date'].dt.hour == 0, 'date'] = hrv_exp[hrv_exp['time_date'].dt.hour == 0]['endDate'].dt.date
hrv_exp.loc[hrv_exp['time_date'].dt.hour > 0, 'date'] = hrv_exp[hrv_exp['time_date'].dt.hour > 0]['startDate'].dt.date
hrv_exp['hours'] = hrv_exp['time_date'].astype('datetime64[s]').dt.time
hrv_exp['new_startDate'] = pd.to_datetime(hrv_exp['date'].astype(str) + ' ' + hrv_exp['hours'].astype(str))
hrv_exp = hrv_exp.drop(['startDate', 'value'], axis=1)
hrv_exp = hrv_exp.rename(columns={'new_startDate': 'startDate', 'bpm': 'value'})
hrv_exp = hrv_exp[['startDate', 'endDate', 'value']]
hrv_exp['value'] = hrv_exp['value'].astype(float)
hrv_exp.reset_index(inplace=True, drop=True)
type = 'HeartRate'
records['is_apple'] = records['sourceName'].str.contains('Apple')
type_records = records.query(f"type == '{type}' and is_apple")
type_records = type_records.drop(['type'], axis=1)
type_records['value'] = type_records['value'].astype(float)
type_records = type_records[['startDate', 'endDate', 'value']]
type_records = pd.concat([type_records, hrv_exp], ignore_index=True)
type_records = type_records[['startDate', 'endDate', 'value']]
return type_records
@staticmethod
def resample_ts(records, resample='5Min', how='mean'):
records = records.sort_values('startDate')
records = records.set_index('startDate')
records = eval(f"records.resample('{resample}').{how}()")
records = records.fillna(method='pad')
records = records.reset_index()
return records
@staticmethod
def trim_outliers(records):
save_records = records.copy()
Q1 = np.percentile(records['value'], 25, interpolation='midpoint')
Q3 = np.percentile(records['value'], 75, interpolation='midpoint')
IQR = Q3 - Q1
upper = Q3 + 1.5 * IQR
lower = Q1 - 1.5 * IQR
records.loc[records['value'] > upper, 'value'] = upper
records.loc[records['value'] < lower, 'value'] = lower
if records['value'].sum() == 0:
return save_records
return records
@staticmethod
def tsfresh(records, prefix):
from tsfresh import extract_features
records = records.sort_values('startDate')
records['timestep'] = records.groupby(['date']).cumcount() + 1
extracted_features = extract_features(records[['date', 'timestep', 'value']], column_id='date', column_sort='timestep')
extracted_features = extracted_features.reset_index()
extracted_features = extracted_features.rename(columns={'index': 'date'})
extracted_features.columns = ['date'] + [prefix + '_' + str(col) for col in extracted_features.columns[1:]]
extracted_features = extracted_features.rename(columns=lambda x: re.sub('[^A-Za-z0-9_]+', '', x))
return extracted_features
@staticmethod
def add_date(records, how):
if how == '24h':
records['date'] = (records['startDate'] - datetime.timedelta(hours=12)).dt.date
elif how == '7days':
records['date'] = records['startDate'].dt.date
df_result = records[0:0]
dates = records['date'].unique()
for d in dates:
min_d = d - datetime.timedelta(days=7)
max_d = d
df_date = records.query('date >= @min_d and date <= @max_d')
df_date['date'] = d
df_result = pd.concat([df_result, df_date])
records = df_result
elif how == '3daysAfter':
records['date'] = records['startDate'].dt.date
df_result = records[0:0]
dates = records['date'].unique()
for d in dates:
min_d = d
max_d = d + datetime.timedelta(days=3)
df_date = records.query('date >= @min_d and date <= @max_d')
df_date['date'] = d
df_result = pd.concat([df_result, df_date])
records = df_result
return records
aggregation_info = {'BoostedHeartRate': 'mean', 'HeartRate': 'mean', 'ActiveEnergyBurned': 'sum', 'StepCount': 'sum', 'BasalEnergyBurned': 'mean', 'AppleStandTime': 'mean', 'FlightsClimbed': 'sum', 'EnvironmentalAudioExposure': 'mean', 'RespiratoryRate': 'mean', 'AppleExerciseTime': 'sum', 'DistanceWalkingRunning': 'sum', 'RespiratoryRate': 'mean', 'HeadphoneAudioExposure': 'mean', 'StairDescentSpeed': 'mean', 'OxygenSaturation': 'mean', 'StairAscentSpeed': 'mean', 'HeartRateVariabilitySDNN': 'mean'}
aggregation_info = {}
for type in tqdm(aggregation_info.keys()):
print(type)
ts_type = FeatureExtractor.get_type_ts(fe.records, type)
prefix = '24h_' + type
ts_resample = FeatureExtractor.resample_ts(ts_type, resample='5Min', how=aggregation_info[type])
ts_resample = FeatureExtractor.trim_outliers(ts_resample)
print(ts_resample.describe())
ts_resample = FeatureExtractor.add_date(ts_resample, how='24h')
ts_resample = FeatureExtractor.tsfresh(ts_resample, prefix)
ts_resample.to_csv(f'{OUTPUT_PATH}/tsfresh_{prefix}.csv', index=False)
prefix = '7d_' + type
ts_resample = FeatureExtractor.resample_ts(ts_type, resample='60min', how=aggregation_info[type])
print(ts_resample.describe())
ts_resample = FeatureExtractor.trim_outliers(ts_resample)
ts_resample = FeatureExtractor.add_date(ts_resample, how='7days')
ts_resample = FeatureExtractor.tsfresh(ts_resample, prefix)
ts_resample.to_csv(f'{OUTPUT_PATH}/tsfresh_{prefix}.csv', index=False)
prefix = '24h_1m_' + type
ts_resample = FeatureExtractor.resample_ts(ts_type, resample='1Min', how='first')
ts_resample = FeatureExtractor.trim_outliers(ts_resample)
print(ts_resample.describe())
ts_resample = FeatureExtractor.add_date(ts_resample, how='24h')
ts_resample = FeatureExtractor.tsfresh(ts_resample, prefix)
ts_resample.to_csv(f'{OUTPUT_PATH}/tsfresh_{prefix}.csv', index=False)
prefix = '24h_1m_10r_' + type
ts_resample = FeatureExtractor.resample_ts(ts_type, resample='1Min', how='first')
ts_resample['value'] = ts_resample['value'].rolling(window=10).var()
ts_resample['value'] = ts_resample['value'].fillna(0)
ts_resample = FeatureExtractor.trim_outliers(ts_resample)
print(ts_resample.describe())
ts_resample = FeatureExtractor.add_date(ts_resample, how='24h')
ts_resample = FeatureExtractor.tsfresh(ts_resample, prefix)
ts_resample.to_csv(f'{OUTPUT_PATH}/tsfresh_{prefix}.csv', index=False) | code |
128018796/cell_14 | [
"text_plain_output_1.png"
] | from boruta import BorutaPy
from os import listdir
from os.path import isfile, join
from pathlib import Path
from sklearn.ensemble import RandomForestRegressor
from tqdm import tqdm
from tsfresh import extract_features
import datetime
import json
import numpy as np
import pandas as pd
import re
import shap
import xmltodict
import datetime
import json
import re
from os import listdir
from os.path import isfile, join
from pathlib import Path
import catboost
import numpy as np
import optuna
import pandas as pd
import shap
import xmltodict
from boruta import BorutaPy
from dateutil.relativedelta import relativedelta
from sklearn.ensemble import RandomForestRegressor
from tqdm import tqdm
pd.options.mode.chained_assignment = None
DATA_ROOT = Path('/kaggle/input/kaggle-pog-series-s01e04/')
DF_TRAIN = DATA_ROOT / 'train.csv'
DF_TEST = DATA_ROOT / 'test.csv'
RAW_DATA = DATA_ROOT / 'raw_health_export.xml'
OUTPUT_PATH = '/kaggle/working/'
TSFRESH_PATH = '/kaggle/input/datasets-sleep/'
train = pd.read_csv(DF_TRAIN)
test = pd.read_csv(DF_TEST)
class FeatureExtractor:
def __init__(self, data_path):
raw_dict = FeatureExtractor.parse_raw_data(data_path)
self.records = FeatureExtractor.get_records_df(raw_dict)
@staticmethod
def parse_raw_data(data_path):
with open(RAW_DATA) as file:
raw_dict = xmltodict.parse(file.read())
return raw_dict
@staticmethod
def get_records_df(raw_dict):
records = pd.DataFrame(raw_dict['HealthData']['Record'])
records.columns = [col.replace('@', '') for col in records.columns]
records['creationDate'] = pd.to_datetime(records['creationDate']).dt.tz_localize(None)
records['startDate'] = pd.to_datetime(records['startDate']).dt.tz_localize(None)
records['endDate'] = pd.to_datetime(records['endDate']).dt.tz_localize(None)
records['type'] = records['type'].str.replace('HKQuantityTypeIdentifier', '')
return records
@staticmethod
def get_type_ts(records, type):
if type != 'BoostedHeartRate':
records['is_apple'] = records['sourceName'].str.contains('Apple')
type_records = records.query(f"type == '{type}' and is_apple")
type_records = type_records.drop(['type'], axis=1)
type_records['value'] = type_records['value'].astype(float)
else:
hrv = records.query('HeartRateVariabilityMetadataList == HeartRateVariabilityMetadataList')
extract_list = lambda x: x['InstantaneousBeatsPerMinute']
hrv['HeartRateVariabilityMetadataList'] = hrv['HeartRateVariabilityMetadataList'].apply(extract_list)
hrv_exp = hrv.explode('HeartRateVariabilityMetadataList')
get_bpm = lambda x: x['@bpm']
get_time = lambda x: x['@time']
hrv_exp['bpm'] = hrv_exp['HeartRateVariabilityMetadataList'].apply(get_bpm)
hrv_exp['time'] = hrv_exp['HeartRateVariabilityMetadataList'].apply(get_time)
hrv_exp['time_date'] = pd.to_datetime(hrv_exp['time'], format='%I:%M:%S.%f %p')
hrv_exp.loc[hrv_exp['time_date'].dt.hour == 0, 'date'] = hrv_exp[hrv_exp['time_date'].dt.hour == 0]['endDate'].dt.date
hrv_exp.loc[hrv_exp['time_date'].dt.hour > 0, 'date'] = hrv_exp[hrv_exp['time_date'].dt.hour > 0]['startDate'].dt.date
hrv_exp['hours'] = hrv_exp['time_date'].astype('datetime64[s]').dt.time
hrv_exp['new_startDate'] = pd.to_datetime(hrv_exp['date'].astype(str) + ' ' + hrv_exp['hours'].astype(str))
hrv_exp = hrv_exp.drop(['startDate', 'value'], axis=1)
hrv_exp = hrv_exp.rename(columns={'new_startDate': 'startDate', 'bpm': 'value'})
hrv_exp = hrv_exp[['startDate', 'endDate', 'value']]
hrv_exp['value'] = hrv_exp['value'].astype(float)
hrv_exp.reset_index(inplace=True, drop=True)
type = 'HeartRate'
records['is_apple'] = records['sourceName'].str.contains('Apple')
type_records = records.query(f"type == '{type}' and is_apple")
type_records = type_records.drop(['type'], axis=1)
type_records['value'] = type_records['value'].astype(float)
type_records = type_records[['startDate', 'endDate', 'value']]
type_records = pd.concat([type_records, hrv_exp], ignore_index=True)
type_records = type_records[['startDate', 'endDate', 'value']]
return type_records
@staticmethod
def resample_ts(records, resample='5Min', how='mean'):
records = records.sort_values('startDate')
records = records.set_index('startDate')
records = eval(f"records.resample('{resample}').{how}()")
records = records.fillna(method='pad')
records = records.reset_index()
return records
@staticmethod
def trim_outliers(records):
save_records = records.copy()
Q1 = np.percentile(records['value'], 25, interpolation='midpoint')
Q3 = np.percentile(records['value'], 75, interpolation='midpoint')
IQR = Q3 - Q1
upper = Q3 + 1.5 * IQR
lower = Q1 - 1.5 * IQR
records.loc[records['value'] > upper, 'value'] = upper
records.loc[records['value'] < lower, 'value'] = lower
if records['value'].sum() == 0:
return save_records
return records
@staticmethod
def tsfresh(records, prefix):
from tsfresh import extract_features
records = records.sort_values('startDate')
records['timestep'] = records.groupby(['date']).cumcount() + 1
extracted_features = extract_features(records[['date', 'timestep', 'value']], column_id='date', column_sort='timestep')
extracted_features = extracted_features.reset_index()
extracted_features = extracted_features.rename(columns={'index': 'date'})
extracted_features.columns = ['date'] + [prefix + '_' + str(col) for col in extracted_features.columns[1:]]
extracted_features = extracted_features.rename(columns=lambda x: re.sub('[^A-Za-z0-9_]+', '', x))
return extracted_features
@staticmethod
def add_date(records, how):
if how == '24h':
records['date'] = (records['startDate'] - datetime.timedelta(hours=12)).dt.date
elif how == '7days':
records['date'] = records['startDate'].dt.date
df_result = records[0:0]
dates = records['date'].unique()
for d in dates:
min_d = d - datetime.timedelta(days=7)
max_d = d
df_date = records.query('date >= @min_d and date <= @max_d')
df_date['date'] = d
df_result = pd.concat([df_result, df_date])
records = df_result
elif how == '3daysAfter':
records['date'] = records['startDate'].dt.date
df_result = records[0:0]
dates = records['date'].unique()
for d in dates:
min_d = d
max_d = d + datetime.timedelta(days=3)
df_date = records.query('date >= @min_d and date <= @max_d')
df_date['date'] = d
df_result = pd.concat([df_result, df_date])
records = df_result
return records
N_FOLDS = 10
RUN_HBO = False
N_TRIALS = 100
N_FEAT = 100
target = 'sleep_hours'
train = pd.read_csv(DF_TRAIN)
test = pd.read_csv(DF_TEST)
def trim_outliers(records):
Q1 = np.percentile(records['sleep_hours'], 25, interpolation='midpoint')
Q3 = np.percentile(records['sleep_hours'], 75, interpolation='midpoint')
IQR = Q3 - Q1
upper = Q3 + 1.5 * IQR
lower = Q1 - 1.5 * IQR
records.loc[records['sleep_hours'] > upper, 'sleep_hours'] = upper
records.loc[records['sleep_hours'] < lower, 'sleep_hours'] = lower
return records
def add_fold(train, N_FOLDS):
np.random.seed(20951)
folds = np.random.randint(0, high=N_FOLDS, size=len(train), dtype=int)
unique, counts = np.unique(folds, return_counts=True)
train['fold'] = folds
dict(zip(unique, counts))
return train
train = add_fold(train, N_FOLDS)
train = train.query("date >= '2020-09-26'").reset_index(drop=True)
def get_feat_boruta(X, y):
model = RandomForestRegressor(n_estimators=100, max_depth=5, random_state=42)
feat_selector = BorutaPy(verbose=0, estimator=model, n_estimators='auto', max_iter=20, two_step=False)
feat_selector.fit(np.array(X), np.array(y))
feats = []
for i in range(len(feat_selector.support_)):
if feat_selector.support_[i]:
feats.append(X.columns[i])
return feats
def global_shap_importance(model, X):
explainer = shap.Explainer(model)
shap_values = explainer(X)
cohorts = {'': shap_values}
cohort_labels = list(cohorts.keys())
cohort_exps = list(cohorts.values())
for i in range(len(cohort_exps)):
if len(cohort_exps[i].shape) == 2:
cohort_exps[i] = cohort_exps[i].abs.mean(0)
features = cohort_exps[0].data
feature_names = cohort_exps[0].feature_names
values = np.array([cohort_exps[i].values for i in range(len(cohort_exps))])
feature_importance = pd.DataFrame(list(zip(feature_names, sum(values))), columns=['features', 'importance'])
feature_importance.sort_values(by=['importance'], ascending=False, inplace=True)
return feature_importance
def get_feat_shap(X, y, n_feats):
model = RandomForestRegressor(n_estimators=20, max_depth=5, random_state=42)
model.fit(X, y)
explainer = shap.TreeExplainer(model)
shap_values = explainer.shap_values(X)
feat_imp = global_shap_importance(model, X)
top_feats = list(feat_imp['features'])[:n_feats]
return top_feats
train_ = train.copy()
test_ = test.copy()
datasets_to_include = [f.split('.')[0] for f in listdir(TSFRESH_PATH) if isfile(join(TSFRESH_PATH, f))]
datasets_to_include = [f for f in datasets_to_include if 'tsfresh' in f]
try:
with open(f'{TSFRESH_PATH}features.json') as json_file:
selection_checkpoint = json.load(json_file)
except:
selection_checkpoint = {}
for d in tqdm(datasets_to_include):
print('Loading:', d)
d_train = train_.copy()
d_train = d_train.merge(pd.read_csv(f'{TSFRESH_PATH}{d}.csv'), on='date', how='left', suffixes=('', f'_{d}'))
d_train = d_train.fillna(0)
d_train = d_train.replace(np.inf, 0)
print('Original Features:', len(d_train.columns) - 3)
X_train = d_train.drop([target, 'date', 'fold'], axis=1)
y_train = d_train[target].tolist()
if d in selection_checkpoint.keys():
selected_feats = selection_checkpoint[d]
else:
selected_feats = get_feat_shap(X_train, y_train, N_FEAT)
X_train = X_train[selected_feats]
selected_feats = get_feat_boruta(X_train, y_train)
d_train = d_train[selected_feats]
d_train['date'] = train_['date']
print('Selected Features:', len(d_train.columns) - 1)
train = train.merge(d_train, on='date', how='left')
train_cols = train.columns.tolist()
train_cols.remove('fold')
d_test = test_.copy()
d_test = d_test.merge(pd.read_csv(f'{TSFRESH_PATH}{d}.csv'), on='date', how='left', suffixes=('', f'_{d}'))
d_test = d_test.fillna(0)
d_test = d_test.replace(np.inf, 0)
d_test = d_test.drop(['sleep_hours'], axis=1)
test = test.merge(d_test, on='date', how='left')[train_cols]
break
train['synthetic'] = False
print('All features num:', len(train.columns)) | code |
33106981/cell_13 | [
"text_plain_output_1.png"
] | from collections import Counter
import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
hotel = pd.read_csv('/kaggle/input/hotel-booking-demand/hotel_bookings.csv')
hotel.shape
hotel.head().T
hotel_num = hotel.dtypes[hotel.dtypes != 'object']
hotel_num = hotel_num.index.to_list()
Date_Drop = {'is_canceled', 'company'}
hotel_num = [ele for ele in hotel_num if ele not in Date_Drop]
hotel_num
hot_num = hotel[hotel_num].copy()
from collections import Counter
def detect_outliers(df, features):
"""
Takes a dataframe df of features and returns a list of the indices
corresponding to the observations containing more than n outliers according
to the Tukey method.
"""
outlier_indices = []
for col in features:
Q1 = np.percentile(df[col], 25)
Q3 = np.percentile(df[col], 75)
IQR = Q3 - Q1
outlier_step = 1.5 * IQR
outlier_list_col = df[(df[col] < Q1 - outlier_step) | (df[col] > Q3 + outlier_step)].index
outlier_indices.extend(outlier_list_col)
outlier_indices = Counter(outlier_indices)
return outlier_indices
Outliers_to_drop = detect_outliers(hotel, hot_num)
len(Outliers_to_drop)
hotel = hotel.drop(Outliers_to_drop, axis=0).reset_index(drop=True)
hotel.isna().sum() | code |
33106981/cell_25 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
hotel = pd.read_csv('/kaggle/input/hotel-booking-demand/hotel_bookings.csv')
hotel.shape
hotel.head().T
hotel_num = hotel.dtypes[hotel.dtypes != 'object']
hotel_num = hotel_num.index.to_list()
Date_Drop = {'is_canceled', 'company'}
hotel_num = [ele for ele in hotel_num if ele not in Date_Drop]
hotel_num
hot_num = hotel[hotel_num].copy()
hotel_num.remove('arrival_date_year')
hotel_num.remove('arrival_date_week_number')
hotel_num.remove('arrival_date_day_of_month')
hotel_num | code |
33106981/cell_23 | [
"text_plain_output_1.png"
] | from collections import Counter
import matplotlib.pyplot as plt
import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import seaborn as sns
hotel = pd.read_csv('/kaggle/input/hotel-booking-demand/hotel_bookings.csv')
hotel.shape
hotel.head().T
hotel_num = hotel.dtypes[hotel.dtypes != 'object']
hotel_num = hotel_num.index.to_list()
Date_Drop = {'is_canceled', 'company'}
hotel_num = [ele for ele in hotel_num if ele not in Date_Drop]
hotel_num
hot_num = hotel[hotel_num].copy()
from collections import Counter
def detect_outliers(df, features):
"""
Takes a dataframe df of features and returns a list of the indices
corresponding to the observations containing more than n outliers according
to the Tukey method.
"""
outlier_indices = []
for col in features:
Q1 = np.percentile(df[col], 25)
Q3 = np.percentile(df[col], 75)
IQR = Q3 - Q1
outlier_step = 1.5 * IQR
outlier_list_col = df[(df[col] < Q1 - outlier_step) | (df[col] > Q3 + outlier_step)].index
outlier_indices.extend(outlier_list_col)
outlier_indices = Counter(outlier_indices)
return outlier_indices
Outliers_to_drop = detect_outliers(hotel, hot_num)
len(Outliers_to_drop)
hotel = hotel.drop(Outliers_to_drop, axis=0).reset_index(drop=True)
hotel.isna().sum()
hotel.company = hotel.company.fillna(0)
hotel.agent = hotel.agent.fillna(0)
hotel.children = hotel.children.fillna(0)
hotel.country = hotel.country.fillna('unknown')
hotel.drop(hotel[(hotel['children'] == 0) & (hotel['babies'] == 0) & (hotel['adults'] == 0)].index, inplace=True)
def cnt_plot(a):
col = hotel[a]
title = 'Category wise count of' + ' ' + a
Cat_Var = hotel.dtypes[hotel.dtypes == 'object']
Cat_Var = Cat_Var.index.to_list()
Date_Drop = {'arrival_date_month', 'reservation_status_date'}
Cat_Var = [ele for ele in Cat_Var if ele not in Date_Drop]
Cat_Var
for col in Cat_Var:
cnt_plot(col) | code |
33106981/cell_6 | [
"text_plain_output_1.png"
] | import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
hotel = pd.read_csv('/kaggle/input/hotel-booking-demand/hotel_bookings.csv')
hotel.shape
hotel.head().T
hotel.describe(percentiles=[0.25, 0.5, 0.75, 0.9, 0.95, 0.99]) | code |
33106981/cell_26 | [
"text_html_output_1.png"
] | from collections import Counter
import matplotlib.pyplot as plt
import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
import seaborn as sns
hotel = pd.read_csv('/kaggle/input/hotel-booking-demand/hotel_bookings.csv')
hotel.shape
hotel.head().T
hotel_num = hotel.dtypes[hotel.dtypes != 'object']
hotel_num = hotel_num.index.to_list()
Date_Drop = {'is_canceled', 'company'}
hotel_num = [ele for ele in hotel_num if ele not in Date_Drop]
hotel_num
hot_num = hotel[hotel_num].copy()
from collections import Counter
def detect_outliers(df, features):
"""
Takes a dataframe df of features and returns a list of the indices
corresponding to the observations containing more than n outliers according
to the Tukey method.
"""
outlier_indices = []
for col in features:
Q1 = np.percentile(df[col], 25)
Q3 = np.percentile(df[col], 75)
IQR = Q3 - Q1
outlier_step = 1.5 * IQR
outlier_list_col = df[(df[col] < Q1 - outlier_step) | (df[col] > Q3 + outlier_step)].index
outlier_indices.extend(outlier_list_col)
outlier_indices = Counter(outlier_indices)
return outlier_indices
Outliers_to_drop = detect_outliers(hotel, hot_num)
len(Outliers_to_drop)
hotel = hotel.drop(Outliers_to_drop, axis=0).reset_index(drop=True)
hotel.isna().sum()
hotel.company = hotel.company.fillna(0)
hotel.agent = hotel.agent.fillna(0)
hotel.children = hotel.children.fillna(0)
hotel.country = hotel.country.fillna('unknown')
hotel.drop(hotel[(hotel['children'] == 0) & (hotel['babies'] == 0) & (hotel['adults'] == 0)].index, inplace=True)
def cnt_plot(a):
col = hotel[a]
title = 'Category wise count of' + ' ' + a
Cat_Var = hotel.dtypes[hotel.dtypes == 'object']
Cat_Var = Cat_Var.index.to_list()
Date_Drop = {'arrival_date_month', 'reservation_status_date'}
Cat_Var = [ele for ele in Cat_Var if ele not in Date_Drop]
Cat_Var
corrmap = hotel.corr()
plt.subplots(figsize=(16, 10))
sns.heatmap(corrmap, annot=True)
plt.show() | code |
33106981/cell_11 | [
"text_plain_output_1.png"
] | from collections import Counter
import numpy as np # linear algebra
import pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)
hotel = pd.read_csv('/kaggle/input/hotel-booking-demand/hotel_bookings.csv')
hotel.shape
hotel.head().T
hotel_num = hotel.dtypes[hotel.dtypes != 'object']
hotel_num = hotel_num.index.to_list()
Date_Drop = {'is_canceled', 'company'}
hotel_num = [ele for ele in hotel_num if ele not in Date_Drop]
hotel_num
hot_num = hotel[hotel_num].copy()
from collections import Counter
def detect_outliers(df, features):
"""
Takes a dataframe df of features and returns a list of the indices
corresponding to the observations containing more than n outliers according
to the Tukey method.
"""
outlier_indices = []
for col in features:
Q1 = np.percentile(df[col], 25)
Q3 = np.percentile(df[col], 75)
IQR = Q3 - Q1
outlier_step = 1.5 * IQR
outlier_list_col = df[(df[col] < Q1 - outlier_step) | (df[col] > Q3 + outlier_step)].index
outlier_indices.extend(outlier_list_col)
outlier_indices = Counter(outlier_indices)
return outlier_indices
Outliers_to_drop = detect_outliers(hotel, hot_num)
len(Outliers_to_drop) | code |
33106981/cell_1 | [
"text_plain_output_1.png"
] | import os
import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
import os
for dirname, _, filenames in os.walk('/kaggle/input'):
for filename in filenames:
print(os.path.join(dirname, filename)) | code |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.