rov-offline/ad_predict_user_data_process.py

import os.path
import time
import datetime
import pandas as pd
from odps import ODPS
from utils import data_check, RedisHelper
from threading import Timer
from config import set_config
config_, _ = set_config()
redis_helper = RedisHelper()

# ODPS服务配置
odps_config = {
    'ENDPOINT': 'http://service.cn.maxcompute.aliyun.com/api',
    'ACCESSID': 'LTAIWYUujJAm7CbH',
    'ACCESSKEY': 'RfSjdiWwED1sGFlsjXv0DlfTnZTG1P',
}

features = [
    'apptype',
    'mid',
    'mid_preview_count_30day',
    'mid_view_count_30day',
    'mid_view_count_pv_30day',
    'mid_play_count_30day',
    'mid_play_count_pv_30day',
    'mid_share_count_30day',
    'mid_share_count_pv_30day',
    'mid_return_count_30day',
    'mid_share_rate_30day',
    'mid_return_rate_30day',
]


def get_feature_data(project, table, dt, app_type):
    """获取特征数据"""
    odps = ODPS(
        access_id=odps_config['ACCESSID'],
        secret_access_key=odps_config['ACCESSKEY'],
        project=project,
        endpoint=odps_config['ENDPOINT'],
    )
    feature_data = []
    sql = f"select * from {project}.{table} where dt={dt} and apptype={app_type} limit 10000"
    with odps.execute_sql(sql).open_reader() as reader:
        for record in reader:
            # print(record)
            item = {}
            for feature_name in features:
                item[feature_name] = record[feature_name]
            feature_data.append(item)
        feature_df = pd.DataFrame(feature_data)
        return feature_df


def user_data_process(project, table, dt, app_type):
    """每日特征处理"""
    print('step 1: get user feature data')
    feature_initial_df = get_feature_data(project=project, table=table, dt=dt, app_type=app_type)
    print(f"feature_initial_df shape: {feature_initial_df.shape}")
    print('step 2: process')
    feature_initial_df['apptype'] = feature_initial_df['apptype'].astype(int)
    feature_df = feature_initial_df.copy()
    # 缺失值填充
    feature_df.fillna(0, inplace=True)
    # 数据类型校正
    type_int_columns = [
        'mid_preview_count_30day',
        'mid_view_count_30day',
        'mid_view_count_pv_30day',
        'mid_play_count_30day',
        'mid_play_count_pv_30day',
        'mid_share_count_30day',
        'mid_share_count_pv_30day',
        'mid_return_count_30day',
    ]
    for column_name in type_int_columns:
        feature_df[column_name] = feature_df[column_name].astype(int)
    type_float_columns = [
        'mid_share_rate_30day',
        'mid_return_rate_30day',
    ]
    for column_name in type_float_columns:
        feature_df[column_name] = feature_df[column_name].astype(float)
    print(f"feature_df shape: {feature_df.shape}")
    print('step 3: add new user feature')
    # 补充新用户默认数据（使用均值）
    new_user_feature = {
        'apptype': app_type,
        'mid': '-1',
        'mid_preview_count_30day': int(feature_df['mid_preview_count_30day'].mean()),
        'mid_view_count_30day': int(feature_df['mid_view_count_30day'].mean()),
        'mid_view_count_pv_30day': int(feature_df['mid_view_count_pv_30day'].mean()),
        'mid_play_count_30day': int(feature_df['mid_play_count_30day'].mean()),
        'mid_play_count_pv_30day': int(feature_df['mid_play_count_pv_30day'].mean()),
        'mid_share_count_30day': int(feature_df['mid_share_count_30day'].mean()),
        'mid_share_count_pv_30day': int(feature_df['mid_share_count_pv_30day'].mean()),
        'mid_return_count_30day': int(feature_df['mid_return_count_30day'].mean()),
    }
    new_user_feature['mid_share_rate_30day'] = float(
        new_user_feature['mid_share_count_pv_30day'] / new_user_feature['mid_play_count_pv_30day'] + 1)
    new_user_feature['mid_return_rate_30day'] = float(
        new_user_feature['mid_return_count_30day'] / new_user_feature['mid_view_count_pv_30day'] + 1)
    new_user_feature_df = pd.DataFrame([new_user_feature])
    user_df = pd.concat([feature_df, new_user_feature_df])
    print(f"user_df shape: {user_df.shape}")
    print(f"step 4: to csv")
    # 写入csv
    predict_data_dir = './data/predict_data'
    if not os.path.exists(predict_data_dir):
        os.makedirs(predict_data_dir)
    user_df.to_csv(f"{predict_data_dir}/user_feature.csv", index=False)
    # to redis
    xgb_config = config_.AD_MODEL_ABTEST_CONFIG['xgb']
    for ind, row in user_df.iterrows():
        app_type = row['apptype']
        mid = row['mid']
        value = {
            'mid_preview_count_30day': row['mid_preview_count_30day'],
            'mid_view_count_30day': row['mid_view_count_30day'],
            'mid_view_count_pv_30day': row['mid_view_count_pv_30day'],
            'mid_play_count_30day': row['mid_play_count_30day'],
            'mid_play_count_pv_30day': row['mid_play_count_pv_30day'],
            'mid_share_count_30day': row['mid_share_count_30day'],
            'mid_share_count_pv_30day': row['mid_share_count_pv_30day'],
            'mid_return_count_30day': row['mid_return_count_30day'],
            'mid_share_rate_30day': row['mid_share_rate_30day'],
            'mid_return_rate_30day': row['mid_return_rate_30day']
        }
        key = f"{xgb_config['predict_user_feature_key_prefix']}{app_type}:{mid}"
        redis_helper.set_data_to_redis(key_name=key, value=str(value), expire_time=48 * 3600)


def timer_check():
    project = 'loghubods'
    table = 'admodel_testset_mid'
    # dt = '20230725'
    now_date = datetime.datetime.today()
    dt = datetime.datetime.strftime(now_date - datetime.timedelta(days=1), '%Y%m%d')
    # 查看当前更新的数据是否已准备好
    data_count = data_check(project=project, table=table, dt=dt)
    if data_count > 0:
        print(f"ad predict user data count = {data_count}")
        # 数据准备好，进行更新
        user_data_process(project=project, table=table, dt=dt, app_type=0)
        print(f"ad predict user data update end!")
    else:
        # 数据没准备好，1分钟后重新检查
        Timer(60, timer_check).start()


if __name__ == '__main__':
    st_time = time.time()
    timer_check()
    print(f"{time.time() - st_time}s")