def create_local_validation():
if DEBUG:
dir_feature = '../processed_features_debug2/'
N = 1503
else:
dir_feature = '../processed_features/'
N = 1503424
filename = dir_feature + 'local_validation.pickle'
if os.path.exists(filename):
local_validation_array = load_pickle(filename)
else:
local_validation_array = [i for i in range(0, N, int(N / 100))]
save_pickle(local_validation_array, filename)
return local_validation_array
# "weight":[7,8,3,8],"price":[4,5,6,6]})
# print(df)
# df['size'] = df.groupby(['fruit','price']).transform(np.size)
# print(df)
# df['freq'] = df.groupby('fruit')['fruit'].transform('count')
# print(df)
# selcols = ['fruit']
# df5 = df[selcols].groupby(selcols).size().reset_index(name="Time4")
# print(df5)
# df = df.merge(df5, on=selcols, how='left')
# print(df)
# feature_name = 'Freq'
# df6 = df[selcols]. \
# groupby(selcols).size(). \
# reset_index(name=feature_name)
# print(df6)
# df = df.merge(df6, on=selcols, how='left')
# print(df)
filename = '../input/aggregated_features.csv'
df = load_csv(filename)
savename = '../input/kernel_aggregated_features.pickle'
save_pickle(df, savename)