rov-offline/cal_hour_score.py

# coding utf-8
import sys
import json
import math
import pandas as pd


features = [
    'apptype',
    'code',
    'videoid',
    'lastonehour_preview',  # 过去1小时预曝光人数
    'lastonehour_view',  # 过去1小时曝光人数
    'lastonehour_play',  # 过去1小时播放人数
    'lastonehour_share',  # 过去1小时分享人数
    'lastonehour_return',  # 过去1小时分享，过去1小时回流人数
    'lastonehour_preview_total',  # 过去1小时预曝光次数
    'lastonehour_view_total',  # 过去1小时曝光次数
    'lastonehour_play_total',  # 过去1小时播放次数
    'lastonehour_share_total',  # 过去1小时分享次数
    'platform_return',
    'lastonehour_show',  # 不区分地域
    'lastonehour_show_region',  # 地域分组
    'lasttwohour_share',  # h-2小时分享人数
    'lasttwohour_return_now',  # h-2分享，过去1小时回流人数
    'lasttwohour_return',  # h-2分享，h-2回流人数
    'lastthreehour_share',  # h-3小时分享人数
    'lastthreehour_return_now',  # h-3分享，过去1小时回流人数
    'lastthreehour_return',  # h-3分享，h-3回流人数

    'lastonehour_return_new',  # 过去1小时分享，过去1小时回流人数（回流统计为对应地域分享带回的回流，分享限制地域，回流不限制地域）
    'lasttwohour_return_now_new',  # h-2分享，过去1小时回流人数（回流统计为对应地域分享带回的回流，分享限制地域，回流不限制地域）
    'lasttwohour_return_new',  # h-2分享，h-2回流人数（回流统计为对应地域分享带回的回流，分享限制地域，回流不限制地域）
    'lastthreehour_return_now_new',  # h-3分享，过去1小时回流人数（回流统计为对应地域分享带回的回流，分享限制地域，回流不限制地域）
    'lastthreehour_return_new',  # h-3分享，h-3回流人数（回流统计为对应地域分享带回的回流，分享限制地域，回流不限制地域）
    'platform_return_new',  # 平台分发回流（回流统计为对应地域分享带回的回流，分享限制地域，回流不限制地域）
]


def data_group(data_path):
    """将数据按照videoid聚合（求和）"""
    f = open(data_path)
    index = 0
    data_dict = {}
    while True:
        line = f.readline()
        if not line:
            break
        if index == 0:
            index += 1
            continue
        index += 1
        items = line.strip().split(",")
        # print(items)
        if len(items) < len(features):
            continue
        video_id = items[2]
        if video_id not in data_dict:
            data_dict[video_id] = {'videoid': video_id}
            for i, feature in enumerate(features):
                if feature in ['apptype', 'code', 'videoid']:
                    continue
                data_dict[video_id][feature] = int(float(items[i]))
        else:
            for i, feature in enumerate(features):
                if feature in ['apptype', 'code', 'videoid']:
                    continue
                data_dict[video_id][feature] = data_dict[video_id][feature] + int(float(items[i]))
    f.close()
    data_list = [item for video_id, item in data_dict.items()]
    data_df = pd.DataFrame(data_list)
    return data_df


def cal_score(data_df):
    """计算score"""
    df = data_df.copy()
    # share_rate_view = (share+1)/(view+1000)
    df['share_rate_view'] = (df['lastonehour_share'] + 1) / (df['lastonehour_view'] + 1000)

    # back_rate = (return+1)/(share+10)
    df['back_rate'] = (df['lastonehour_return'] + 1) / (df['lastonehour_share'] + 10)
    # back_rate_2h = (lasttwohour_return_now+1)/(share+10)
    df['back_rate_2h'] = (df['lasttwohour_return_now'] + 1) / (df['lasttwohour_share'] + 10)
    # back_rate_3h = (lastthreehour_return_now+1)/(share+10)
    df['back_rate_3h'] = (df['lastthreehour_return_now'] + 1) / (df['lastthreehour_share'] + 10)

    df['log_back'] = (df['lastonehour_return'] + 1).apply(math.log)

    # h-2回流留存
    df['return_retention_initial_2h'] = (df['lasttwohour_return_now'] + 1) / (df['lasttwohour_return'] + 5)
    df['return_retention_2h'] = df['return_retention_initial_2h'].apply(lambda x: 1 if x > 1 else x)
    # h-3回流留存
    df['return_retention_initial_3h'] = (df['lastthreehour_return_now'] + 1) / (df['lastthreehour_return'] + 10)
    df['return_retention_3h'] = df['return_retention_initial_3h'].apply(lambda x: 0.8 if x > 0.8 else x)

    # score1 = 回流/(view+5)
    df['hour_score1'] = df['lastonehour_return'] / (df['lastonehour_view'] + 5)

    # score2 = (回流 * (1 + h-2回流留存 + h-3回流留存))/(view+1000)
    df['hour_score2'] = (df['lastonehour_return'] * (1 + df['return_retention_2h'] + df['return_retention_3h'])) / \
                   (df['lastonehour_view'] + 1000)

    # score3 = (lastthreehour_return_now + lasttwohour_return_now + lastonehour_return)/(lastonehour_view+1000)
    df['hour_score3'] = (df['lastthreehour_return_now'] + df['lasttwohour_return_now'] + df['lastonehour_return']) / \
                   (df['lastonehour_view'] + 1000)

    # score4 = share/view * back_rate * logback
    df['hour_score4'] = df['share_rate_view'] * df['back_rate'] * df['log_back']

    # score5 = share/view * (back_rate + back_rate_2h + back_rate_3h) * logback
    df['hour_score5'] = df['share_rate_view'] * (df['back_rate'] + df['back_rate_2h'] + df['back_rate_3h']) * df['log_back']

    score_df = df[['videoid', 'hour_score1', 'hour_score2', 'hour_score3', 'hour_score4', 'hour_score5']]
    # print(score_df)
    return score_df


if __name__ == "__main__":
    # 1.load data
    now_date = sys.argv[1]
    print(f"now_date: {now_date}")
    data_path = f"./data/hour_video_data_{now_date}.csv"
    data_df = data_group(data_path=data_path)
    print(f"data_df shape: {data_df.shape}")
    hour_score_path = f"./data/hour_score_{now_date}.csv"
    score_df = cal_score(data_df=data_df)
    score_df.to_csv(hour_score_path, index=False)
    print(f"score_df shape: {score_df.shape}")