from conf.configure import Configure
from reading.advertisement import Advertisement
from reading.app import App
from reading.dataset import Dataset
from reading.position import Position
from reading.user import User
from reading.user_app_actions import User_App_Actions
import matplotlib.pyplot as plt
import numpy as np
if __name__ == '__main__':
debug = False
print('reading data...')
ad = Advertisement(Configure.ad_path, debug=debug)
app = App(Configure.app_categories_path, debug=debug)
data_set = Dataset(Configure.train_path, Configure.test_path, debug=debug)
position = Position(Configure.position_path, debug=debug)
user = User(Configure.user_path, debug=debug)
user_app_actions = User_App_Actions(Configure.user_app_actions_path, debug=debug)
days = [0] * 32
hours = [0] * 24
minutes = [0] * 60
conver_days = [0] * 32
conver_hours = [0] * 24
conver_minutes = [0] * 60
click_in_con_days= [0] * 32
click_in_con_hours = [0] * 24
import pandas as pd
from conf.configure import Configure
from reading.advertisement import Advertisement
from reading.dataset import Dataset
from reading.user import User
from reading.app import App
if __name__ == '__main__':
ad = Advertisement(Configure.ad_path)
app = App(Configure.app_categories_path)
data_set = Dataset(Configure.train_path, Configure.test_path)
user = User(Configure.user_path)
result0 = [0] * 100
result1 = [0] * 100
for d in data_set.get_data_list():
userID = d['userID']
age = user.get_value(userID)['age']
label = d['label']
if label < 0:
continue
if label == 0:
result0[age] += 1
else:
result1[age] += 1
for i in range(len(result0)):
print(i, result0[i] + result1[i], result0[i], result1[i])
def add_to_app_cat(self, ad: Advertisement, app: App):
for record in self.data_list:
creativeID = record.creativeID
appID = ad.get_value(creativeID).appID
appCategory = app.get_value(appID).appCategory
app.add_dataset(record, appCategory)
if __name__ == '__main__':
# ad = pd.read_csv(Configure.ad_path)
# data_set = pd.read_csv(Configure.train_path)
# user = pd.read_csv(Configure.user_path)
# app = pd.read_csv(Configure.app_categories_path)
#
# data = data_set.merge(user, how='left', on='userID')
# data = data.merge(ad, how='left', on='creativeID')
# data = data.merge(app, how='left', on='appID')
# data = data[data['label'] >= 0]
# g = data.groupby(['appCategory', 'haveBaby', 'label']).count()
# for group_name, df in g:
# print(group_name, df)
ad = Advertisement(Configure.ad_path)
app = App(Configure.app_categories_path)
data_set = Dataset(Configure.train_path, Configure.test_path)
user = User(Configure.user_path)
from_t = 270000
end_t = 290000
result = dict()
for d in data_set.get_data_list():
creativeID = d['creativeID']
userID = d['userID']
haveBaby = user.get_value(userID)['haveBaby']
appID = ad.get_value(creativeID)['appID']
appCategory = app.get_value(appID)['appCategory']
label = d['label']
clickTime = d['clickTime']
if clickTime < from_t or clickTime >= end_t: