恢复策略

alg_growth_3rd_gh_reply_video_v1.py

@@ -23,227 +23,139 @@ BASE_GROUP_NAME = '3rd-party-base'
 EXPLORE1_GROUP_NAME = '3rd-party-explore1'
 EXPLORE2_GROUP_NAME = '3rd-party-explore2'
 #TODO: fetch gh_id from external data source
-GH_IDS = ('gh_2863155cedcb',
-'gh_c1acd6bac0f8',
-'gh_da993c5f7f64',
-'gh_495d71abbda2',
-'gh_e2318164f869',
-'gh_fc4ec610756e',
-'gh_2450ad774945',
-'gh_175925e40318',
-'gh_994adaf7a539',
-'gh_250c51d5ce69',
-'gh_37adcd637351',
-'gh_a36405f4e5d3',
-'gh_ee5b4b07ed8b',
-'gh_11debb2c8392',
-'gh_d645c1ef7fb0',
-'gh_1899b728af86',
-'gh_059a27ea86b2',
-'gh_6454c103be14',
-'gh_63745bad4f21',
-'gh_8a29eebc2012',
-'gh_57bc9846c86a',
-'gh_570967881eae',
-'gh_197a0d8caa31',
-'gh_93af434e3f47',
-'gh_184f2d765f55',
-'gh_8157d8fd284e',
-'gh_8e1d1f19d44f',
-'gh_1da8f62f4a0d',
-'gh_fd4df7c45bb9',
-'gh_dcfcf74b0846',
-'gh_3afc3a8b8a3d',
-'gh_ef699270bf64',
-'gh_ba870f8b178b',
-'gh_58cdb2f1f0d0',
-'gh_3dce33de6994',
-'gh_543e6d7e15f3',
-'gh_0d55fee6b78d',
-'gh_1c11651e0df4',
-'gh_7f4741dd5fea',
-'gh_33e9e4cbed84',
-'gh_23fa67ebb016',
-'gh_33c26df4eab0',
+GH_IDS = (
 'gh_01c93b07605f',
-'gh_c655e3c8a121',
-'gh_83adb78f4ede',
+'gh_041d1c819c30',
+'gh_04804a94e325',
+'gh_059a27ea86b2',
+'gh_06699758fa4b',
 'gh_0cc7c6d712eb',
-'gh_1b8bfb5c4ffd',
-'gh_e4fb77b1023b',
-'gh_f994f5d9a4b6',
-'gh_e0ca8ba4ed91',
-'gh_b2b4d5aa6b49',
-'gh_53759f90b0c5',
-'gh_d219a0cc8a35',
-'gh_930b5ef5a185',
-'gh_22cad14dc4ec',
-'gh_8734c02f2983',
-'gh_8d68e68f2d08',
-'gh_c603033bf881',
-'gh_55ac4e447179',
-'gh_8b5c838ac19a',
-'gh_aed71f26e7e6',
-'gh_330ef0db846d',
-'gh_87eca527c626',
-'gh_7a14b4c15090',
-'gh_b74693fed783',
-'gh_e1594e6db64b',
-'gh_d32daba8ccf8',
-'gh_23e084923def',
+'gh_0d3c97cc30cc',
+'gh_11debb2c8392',
+'gh_126c99b39cea',
+'gh_133c36b99b14',
+'gh_143743361496',
 'gh_148aa4a776ce',
-'gh_0df4d6b647ea',
-'gh_041d1c819c30',
-'gh_7e33fbba4398',
-'gh_354ab82cf9b3',
-'gh_b1f1d7a1f351',
-'gh_793647539ef5',
+'gh_184f2d765f55',
+'gh_197a0d8caa31',
+'gh_1b243f162cbd',
+'gh_1b8bfb5c4ffd',
+'gh_1c11651e0df4',
+'gh_1cbb453a800e',
+'gh_1da8f62f4a0d',
 'gh_1ff0c29f8408',
-'gh_ecef1c08bcf4',
+'gh_210f7ce6f418',
+'gh_22cad14dc4ec',
 'gh_22f53f6a2b5d',
+'gh_23e084923def',
+'gh_23fa67ebb016',
+'gh_2450ad774945',
+'gh_250c51d5ce69',
+'gh_261bbd99a906',
+'gh_26f1353fda5a',
+'gh_2863155cedcb',
+'gh_2fbff340c683',
+'gh_30b472377707',
+'gh_330ef0db846d',
+'gh_33731afddbdb',
+'gh_33c26df4eab0',
+'gh_33e9e4cbed84',
 'gh_34820675d0fc',
+'gh_354ab82cf9b3',
+'gh_36e74017026e',
+'gh_37adcd637351',
+'gh_3afc3a8b8a3d',
+'gh_3dce33de6994',
+'gh_3ee85ba7c3ae',
+'gh_40ff43c50773',
 'gh_4175a8f745f6',
-'gh_81145598368a',
-'gh_5f0bb5822e10',
-'gh_65d8db4e97ca',
-'gh_a09594d52fda',
-'gh_4411cf1e5f4e',
-'gh_9ee5f5e8425f',
-'gh_df24adad2521',
-'gh_30b472377707',
-'gh_bb6775b47656',
-'gh_69808935bba0',
-'gh_fb77872bf907',
-'gh_830c4aa1b262',
-'gh_b5393e35caa4',
-'gh_fa7dceae7c9d',
 'gh_449fb0c2d817',
-'gh_d6e75ad9094f',
-'gh_1cbb453a800e',
-'gh_1b243f162cbd',
-'gh_50db6881c86e',
-'gh_9d94775e8137',
-'gh_d37101fb9b98',
-'gh_ed86d05703eb',
-'gh_ac4072121e24',
-'gh_620af8e24fb9',
-'gh_ee4783ded544',
-'gh_d2bb5f1b9498',
-'gh_5044de6e1597',
-'gh_d94de77a8d08',
-'gh_98624814f69a',
+'gh_45be25d1c06b',
+'gh_491189a534f2',
+'gh_495d71abbda2',
 'gh_4c38b9d4474a',
-'gh_f2a6c90c56cb',
-'gh_26f1353fda5a',
-'gh_143743361496',
-'gh_126c99b39cea',
-'gh_53e6e0a1b1bd',
-'gh_859aafbcda3d',
-'gh_cfce2617bd82',
-'gh_db8ea2bc6687',
-'gh_c4708b8cfe39',
+'gh_4d2f75c3c3fe',
+'gh_5044de6e1597',
+'gh_53759f90b0c5',
+'gh_543e6d7e15f3',
+'gh_5522900b6a67',
+'gh_55ac4e447179',
+'gh_56b65b7d4520',
+'gh_570967881eae',
+'gh_57bc9846c86a',
 'gh_57d2388bd01d',
+'gh_58cdb2f1f0d0',
+'gh_5e0cd3f7b457',
+'gh_5f0bb5822e10',
 'gh_5fffe35cc12a',
-'gh_45980a6448f3',
-'gh_f5120c12ee23',
-'gh_bf79e3645d7a',
-'gh_6c6d81dd642d',
-'gh_57ee6a6ef204',
-'gh_45be25d1c06b',
-'gh_3ee85ba7c3ae',
-'gh_7c89d5a3e745',
-'gh_c46be9ea4eef',
-'gh_cedc3c4eb48b',
-'gh_8a91fa7f32aa',
-'gh_5207b355776f',
+'gh_63745bad4f21',
+'gh_6454c103be14',
+'gh_69808935bba0',
 'gh_6c7f73de400b',
-'gh_d2f3805f8fa3',
-'gh_7dd47f8aca4e',
-'gh_967f9abb9ccd',
-'gh_f46c6c9b53fa',
-'gh_086abf2a536b',
-'gh_6e11282216f3',
-'gh_f5332b8dfb63',
-'gh_f78610e292ba',
-'gh_06699758fa4b',
-'gh_92323d0bea11',
-'gh_517aed4e8197',
-'gh_c80462b5a330',
-'gh_1b1c3ced734e',
-'gh_dd54e30b03ad',
-'gh_cadd0ea4fab3',
-'gh_ef07a709127e',
-'gh_ab6ca922e605',
-'gh_8b69b67ea723',
-'gh_363c54315788',
-'gh_a363987c60bf',
-'gh_86ca35774fcf',
-'gh_518694803ae7',
-'gh_f98d5f17e9ea',
-'gh_5e0cd3f7b457',
-'gh_9e0d149e2c0a',
+'gh_7a14b4c15090',
+'gh_7c89d5a3e745',
+'gh_7e33fbba4398',
 'gh_7e77b09bb4f5',
-'gh_261bbd99a906',
-'gh_2dc8e3a7b6c9',
-'gh_1ec8dae66c97',
 'gh_7f062810b4e7',
-'gh_3c112c0c9c8b',
-'gh_01cd19465b39',
-'gh_8cc8ae6eb9a5',
-'gh_210f7ce6f418',
-'gh_04804a94e325',
-'gh_4685665647f0',
-'gh_d7fa96aeb839',
-'gh_210cb680d83d',
-'gh_862b00a394e3',
-'gh_3cf7b310906a',
-'gh_669555ebea28',
-'gh_aaac62205137',
-'gh_0a03f8fa63ba',
-'gh_b8b2d4184832',
-'gh_819a632d4bb1',
-'gh_db09b87a0fc9',
-'gh_b673c01e7bd8',
-'gh_6da61a15044a',
-'gh_2f1fab4efaef',
-'gh_da22f64152d5',
-'gh_ff9fe99f2097',
-'gh_33731afddbdb',
-'gh_4d2f75c3c3fe',
-'gh_40ff43c50773',
-'gh_56b65b7d4520',
-'gh_ff16c412ab97',
+'gh_8157d8fd284e',
+'gh_83adb78f4ede',
+'gh_859aafbcda3d',
+'gh_86ca35774fcf',
+'gh_87eca527c626',
+'gh_8a29eebc2012',
+'gh_8b5c838ac19a',
+'gh_8b69b67ea723',
 'gh_8bf689ae15cc',
-'gh_650b17dbba8f',
-'gh_b63b9dde3f4b',
-'gh_36e74017026e',
+'gh_8cc8ae6eb9a5',
+'gh_8d68e68f2d08',
+'gh_930b5ef5a185',
+'gh_93af434e3f47',
+'gh_98624814f69a',
+'gh_9d94775e8137',
+'gh_9e0c7a370aaf',
+'gh_9e0d149e2c0a',
+'gh_9ee5f5e8425f',
+'gh_a09594d52fda',
+'gh_a36405f4e5d3',
 'gh_a8851bfa953b',
-'gh_ec5beb465640',
-'gh_133c36b99b14',
-'gh_b144210318e5',
-'gh_3bffce62dbb4',
-'gh_2fbff340c683',
-'gh_3ceae370dcf5',
-'gh_530b634707b0',
+'gh_ac4072121e24',
+'gh_aed71f26e7e6',
+'gh_b63b9dde3f4b',
 'gh_b7cdece20099',
-'gh_9e0c7a370aaf',
-'gh_96412c0393e3',
+'gh_ba870f8b178b',
+'gh_bb6775b47656',
+'gh_bf79e3645d7a',
+'gh_c1acd6bac0f8',
+'gh_c4708b8cfe39',
+'gh_c603033bf881',
 'gh_c8060587e6d1',
-'gh_0d3c97cc30cc',
-'gh_491189a534f2',
-'gh_fe9620386c2c',
-'gh_9d50b7067f07',
-'gh_e1331141406a',
-'gh_d6db13fcf14d',
-'gh_5522900b6a67',
-'gh_a7c21403c493',
-'gh_eeec7c2e28a5',
-'gh_c783350a9660',)
+'gh_cedc3c4eb48b',
+'gh_d219a0cc8a35',
+'gh_d2bb5f1b9498',
+'gh_d32daba8ccf8',
+'gh_d6e75ad9094f',
+'gh_dd54e30b03ad',
+'gh_e0ca8ba4ed91',
+'gh_e2318164f869',
+'gh_e4fb77b1023b',
+'gh_ecef1c08bcf4',
+'gh_ee5b4b07ed8b',
+'gh_ef699270bf64',
+'gh_f2a6c90c56cb',
+'gh_f46c6c9b53fa',
+'gh_f5120c12ee23',
+'gh_fb77872bf907',
+'gh_fc4ec610756e',
+'gh_ff16c412ab97',
+'gh_ff9fe99f2097',
+)
 CDN_IMG_OPERATOR = "?x-oss-process=image/resize,m_fill,w_600,h_480,limit_0/format,jpg/watermark,image_eXNoL3BpYy93YXRlcm1hcmtlci9pY29uX3BsYXlfd2hpdGUucG5nP3gtb3NzLXByb2Nlc3M9aW1hZ2UvcmVzaXplLHdfMTQ0,g_center"
 
 ODS_PROJECT = "loghubods"
+EXPLORE_POOL_TABLE = 'alg_growth_video_return_stats_history'
 GH_REPLY_STATS_TABLE = 'alg_growth_3rd_gh_reply_video_stats'
-ODPS_RANK_RESULT_TABLE = 'alg_gh_autoreply_video_rank_data'
+# ODPS_RANK_RESULT_TABLE = 'alg_gh_autoreply_video_rank_data'
+ODPS_3RD_RANK_RESULT_TABLE = 'alg_3rd_gh_autoreply_video_rank_data'
 RDS_RANK_RESULT_TABLE = 'alg_gh_autoreply_video_rank_data'
 STATS_PERIOD_DAYS = 5
 SEND_N = 1
@@ -298,6 +210,73 @@ def process_reply_stats(project, table, period, run_dt):
     merged_df['score'] = merged_df['day0_return'] / (merged_df['send_count'] + 500)
     return merged_df
 
+def rank_for_layer1(run_dt, run_hour, project, table):
+    # TODO: 加审核&退场
+    df = get_odps_df_of_max_partition(project, table, {'dt': run_dt})
+    df = df.to_pandas()
+    # 确保重跑时可获得一致结果
+    dt_version = f'{run_dt}{run_hour}'
+    np.random.seed(int(dt_version)+1)
+
+    # TODO: 修改权重计算策略
+    df['score'] = df['rov']
+
+    sampled_df = df.sample(n=SEND_N, weights=df['score'])
+    sampled_df['sort'] = range(1, len(sampled_df) + 1)
+    sampled_df['strategy_key'] = EXPLORE1_GROUP_NAME
+    sampled_df['dt_version'] = dt_version
+
+    gh_name_df = pd.DataFrame({'gh_id': GH_IDS + ('default', )})
+    sampled_df['_tmpkey'] = 1
+    gh_name_df['_tmpkey'] = 1
+    extend_df = sampled_df.merge(gh_name_df, on='_tmpkey').drop('_tmpkey', axis=1)
+
+    result_df = extend_df[['strategy_key', 'dt_version', 'gh_id', 'sort', 'video_id', 'score']]
+    return result_df
+
+def rank_for_layer2(run_dt, run_hour, project, stats_table, rank_table):
+    stats_df = process_reply_stats(project, stats_table, STATS_PERIOD_DAYS, run_dt)
+
+    # 确保重跑时可获得一致结果
+    dt_version = f'{run_dt}{run_hour}'
+    np.random.seed(int(dt_version)+1)
+    # TODO: 计算账号间相关性
+    ## 账号两两组合，取有RoVn数值视频的交集，单个账号内的RoVn（平滑后）组成向量
+    ## 求向量相关系数或cosine相似度
+    ## 单个视频的RoVn加权求和
+    # 当前实现基础版本：只在账号内求二级探索排序分
+
+    sampled_dfs = []
+    # 处理default逻辑（default-explore2）
+    default_stats_df = stats_df.query('gh_id == "default"')
+    sampled_df = default_stats_df.sample(n=SEND_N, weights=default_stats_df['score'])
+    sampled_df['sort'] = range(1, len(sampled_df) + 1)
+    sampled_dfs.append(sampled_df)
+
+    # 基础过滤for账号
+    df = stats_df.query('day0_return > 100')
+
+    # fallback to base if necessary
+    base_strategy_df = get_last_strategy_result(
+        project, rank_table, dt_version, BASE_GROUP_NAME)
+
+    for gh_id in GH_IDS:
+        sub_df = df.query(f'gh_id == "{gh_id}"')
+        if len(sub_df) < SEND_N:
+            LOGGER.warning(
+                "gh_id[{}] rows[{}] not enough for layer2, fallback to base"
+                .format(gh_id, len(sub_df)))
+            sub_df = base_strategy_df.query(f'gh_id == "{gh_id}"')
+            sub_df['score'] = sub_df['sort']
+        sampled_df = sub_df.sample(n=SEND_N, weights=sub_df['score'])
+        sampled_df['sort'] = range(1, len(sampled_df) + 1)
+        sampled_dfs.append(sampled_df)
+
+    extend_df = pd.concat(sampled_dfs)
+    extend_df['strategy_key'] = EXPLORE2_GROUP_NAME
+    extend_df['dt_version'] = dt_version
+    result_df = extend_df[['strategy_key', 'dt_version', 'gh_id', 'sort', 'video_id', 'score']]
+    return result_df
 
 def rank_for_base(run_dt, run_hour, project, stats_table, rank_table, stg_key):
     stats_df = process_reply_stats(project, stats_table, STATS_PERIOD_DAYS, run_dt)
@@ -344,17 +323,17 @@ def check_result_data(df):
     for gh_id in GH_IDS + ('default', ):
         for key in (EXPLORE1_GROUP_NAME, EXPLORE2_GROUP_NAME, BASE_GROUP_NAME):
             sub_df = df.query(f'gh_id == "{gh_id}" and strategy_key == "{key}"')
-            # if len(sub_df) != SEND_N:
-            #     raise Exception(f"Result not enough for gh_id[{gh_id}]")
+            if len(sub_df) != SEND_N:
+                raise Exception(f"Result not enough for gh_id[{gh_id}]")
 
 
 def build_and_transfer_data(run_dt, run_hour, project, **kwargs):
     dt_version = f'{run_dt}{run_hour}'
     dry_run = kwargs.get('dry_run', False)
 
-    layer1_rank = rank_for_base(run_dt, run_hour, ODS_PROJECT, GH_REPLY_STATS_TABLE, ODPS_RANK_RESULT_TABLE, EXPLORE1_GROUP_NAME)
-    layer2_rank = rank_for_base(run_dt, run_hour, ODS_PROJECT, GH_REPLY_STATS_TABLE, ODPS_RANK_RESULT_TABLE, EXPLORE2_GROUP_NAME)
-    base_rank = rank_for_base(run_dt, run_hour, ODS_PROJECT, GH_REPLY_STATS_TABLE, ODPS_RANK_RESULT_TABLE, BASE_GROUP_NAME)
+    layer1_rank = rank_for_layer1(run_dt, run_hour, ODS_PROJECT, EXPLORE_POOL_TABLE)
+    layer2_rank = rank_for_layer2(run_dt, run_hour, ODS_PROJECT, GH_REPLY_STATS_TABLE, ODPS_3RD_RANK_RESULT_TABLE)
+    base_rank = rank_for_base(run_dt, run_hour, ODS_PROJECT, GH_REPLY_STATS_TABLE, ODPS_3RD_RANK_RESULT_TABLE, BASE_GROUP_NAME)
     final_rank_df = pd.concat([layer1_rank, layer2_rank, base_rank]).reset_index(drop=True)
     check_result_data(final_rank_df)
 
@@ -377,11 +356,11 @@ def build_and_transfer_data(run_dt, run_hour, project, **kwargs):
         return
 
     # save to ODPS
-    # t = odps_instance.get_table(ODPS_RANK_RESULT_TABLE)
-    # part_spec_dict = {'dt': run_dt, 'hour': run_hour, 'ctime': dt_version}
-    # part_spec =','.join(['{}={}'.format(k, part_spec_dict[k]) for k in part_spec_dict.keys()])
-    # with t.open_writer(partition=part_spec, create_partition=True, overwrite=True) as writer:
-    #     writer.write(list(final_df.itertuples(index=False)))
+    t = odps_instance.get_table(ODPS_3RD_RANK_RESULT_TABLE)
+    part_spec_dict = {'dt': run_dt, 'hour': run_hour, 'ctime': dt_version}
+    part_spec =','.join(['{}={}'.format(k, part_spec_dict[k]) for k in part_spec_dict.keys()])
+    with t.open_writer(partition=part_spec, create_partition=True, overwrite=True) as writer:
+        writer.write(list(final_df.itertuples(index=False)))
 
     # sync to MySQL
     data_to_insert = [tuple(row) for row in final_df.itertuples(index=False)]