def main():
df = pd.read_csv('train_format2.csv', index_col='user_id')
specific_user_tensor_info = df.ix[34176, ['merchant_id', 'activity_log']]
specific_user_tensor = Tensor(specific_user_tensor_info)
specific_user_tensor.build_tensor()
one_click_list = specific_user_tensor.is_one_click()
user_df = df.ix[34176]
user_df = user_df.reset_index()
user_df = user_df.drop(one_click_list)
user_df.to_csv('34176_clean_data.csv', index=False)
def transform_csv_to_liner_tensor(df):
df.fillna(0)
user_id_list = list(deque(df.index.values))
users_tensor = dict()
# help debug
user_time = 0
user_count = len(user_id_list)
print "total user_count:", user_count
for user_id in user_id_list:
print "the times:", user_time,
print "the user_id:", user_id,
user_tensor_info = df.ix[user_id, ['merchant_id', 'activity_log']]
specific_user_tensor = Tensor(user_tensor_info)
specific_user_tensor.build_tensor()
users_tensor[user_id] = specific_user_tensor
user_time = user_time + 1
print "percent:", user_time/user_count
return users_tensor
