| product_tag: 'lookalike' |
| pipeline_tag: '03042021' |
| persona_table_name: 'ads_persona_0520' |
| showlog_table_name: 'ads_showlog_0520' |
| clicklog_table_name: 'ads_clicklog_0520' |
| keywords_table: 'din_ad_keywords_02182021' #if this keywords table name is changed, set the below to True. |
| log: |
| level: 'WARN' # log level for spark and app |
| features: |
| features: [ |
| 'slot_id', |
| 'inter_type_cd', |
| 'creat_type_cd', |
| # 'app_name', |
| 'package_name', |
| 'adv_device_type_cd', |
| 'os_ver', |
| 'device_name', |
| 'os_language', |
| 'req_date', |
| 'media_busin_id', |
| 'plat_income_amt', |
| 'site_id', |
| 'device_size', |
| 'slot_status_cd', |
| 'sdk_ver', |
| 'screen_resolution', |
| 'net_type', |
| 'carrier', |
| 'req_time', |
| 'return_time', |
| 'show_id', |
| 'show_time', |
| 'inside_channel_id', |
| 'bill_env', |
| 'media_busin_income_amt', |
| 'adv_prim_class_cd', |
| 'ssp_dsp_channel_id', |
| 'adv_bill_mode_cd', |
| 'oper_cnt', |
| 'etl_time', |
| 'show_freq_ctrl_mode', |
| 'client_report_event_id', |
| 'record_time_msec', |
| 'repeat_report_count', |
| 'last_report_fail_reason', |
| 'position_id', |
| 'award_type_cd', |
| 'adx_order_id', |
| 'content_down_method_cd', |
| 'trade_type', |
| 'spread_app_id', |
| 'spread_app_industry_id', |
| 'spread_daily_budget', |
| 'adx_deal_cd', |
| 'device_id_type_cd', |
| 'oaid', |
| 'enable_track_flg', |
| 'valid_cd', |
| 'max_adv_count', |
| 'evt_jump_src_cd', |
| 'evt_jump_src_fail_cd', |
| 'chk_result', |
| 'second_agent_id', |
| 'adv_valid_period', |
| 'first_agent_id', |
| 'dynamic_creat_template_id', |
| 'goods_lbry_id', |
| 'goods_id', |
| 'indu_type', |
| 'click_success_jump_dst', |
| 'evt_uuid', |
| 'pay_type_flg', |
| 'cuerrency', |
| 'cfolder_type_flg', |
| 'return_recommend_val', |
| 'adx_order_purch_cnt', |
| 'total_budget', |
| 'cp_id', |
| 'media_customize_acct_par1', |
| 'teleplay_id', |
| 'exp_cnt', |
| 'exp_src', |
| 'adv_show_contin_duration', |
| 'max_show_ratio', |
| 'adx_order_total_purch_cnt', |
| 'media_receipt_acct', |
| 'media_receipt_currency', |
| 'service_judge_invalid_rule_code', |
| 'realtm_charge_result_rule_code', |
| 'plat_actual_cash_income', |
| 'daily_show_max_cnt', |
| 'slot_size', |
| 'test_adv_flg', |
| 'app_ver', |
| 'adv_type', |
| 'adv_prim_id', |
| 'order_id', |
| 'task_id', |
| 'pps_inside_exprmt_ab_tag'] |
| pipeline: |
| main_clean: |
| did_bucket_num: 2 # Number of partitions for did |
| load_logs_in_minutes: 14400 #1440/day, original=14400 |
| create_keywords: False # set True for first run, then keep False to use the created table. |
| persona_output_table: '{product_tag}_{pipeline_tag}_persona' |
| showlog_output_table: '{product_tag}_{pipeline_tag}_showlog' |
| clicklog_output_table: '{product_tag}_{pipeline_tag}_clicklog' |
| conditions: { |
| 'new_slot_id_list': [ |
| '06', |
| '11', |
| '05', |
| '04', |
| '03', |
| '02', |
| '01', |
| 'l03493p0r3', |
| 'x0ej5xhk60kjwq', |
| 'g7m2zuits8', |
| 'w3wx3nv9ow5i97', |
| 'a1nvkhk62q', |
| 'g9iv6p4sjy', |
| 'd47737w664', |
| 'c4n08ku47t', |
| 'b6le0s4qo8', |
| 'd9jucwkpr3', |
| 'p7gsrebd4m', |
| 'a8syykhszz', |
| 'l2d4ec6csv', |
| 'j1430itab9wj3b', |
| 'h034y5sp0i', |
| 's4z85pd1h8', |
| 'z041bf6g4s', |
| '71bcd2720e5011e79bc8fa163e05184e', |
| 'a47eavw7ex', |
| '68bcd2720e5011e79bc8fa163e05184e', |
| '66bcd2720e5011e79bc8fa163e05184e', |
| '72bcd2720e5011e79bc8fa163e05184e', |
| 'f1iprgyl13', |
| 'q4jtehrqn2', |
| 'm1040xexan', |
| 'd971z9825e', |
| 'a83jryvehg', |
| '7b0d7b55ab0c11e68b7900163e3e481d', |
| 'a290af82884e11e5bdec00163e291137', |
| 'w9fmyd5r0i', |
| 'x2fpfbm8rt', |
| 'e351de37263311e6af7500163e291137', |
| 'k4werqx13k', |
| '5cd1c663263511e6af7500163e291137', |
| '17dd6d8098bf11e5bdec00163e291137', |
| 'd4d7362e879511e5bdec00163e291137', |
| '15e9ddce941b11e5bdec00163e291137' |
| ], |
| 'starting_date': '2019-12-18', #2019-12-18 |
| 'ending_date': '2020-04-16' #2020-04-15 #2019-12-23 |
| } |
| main_logs: |
| interval_time_in_seconds: 86400 # default=1 day, group logs in interval time. |
| logs_output_table_name: '{product_tag}_{pipeline_tag}_logs' |
| main_trainready: |
| trainready_output_table: '{product_tag}_{pipeline_tag}_trainready' |
| tfrecords: |
| tfrecords_statistics_path: '{product_tag}_{pipeline_tag}_tfrecord_statistics.pkl' |
| tfrecords_hdfs_path: '{product_tag}_{pipeline_tag}_tfrecord' # it is hdfs location for tfrecords, over-writes the existing files |
| cutting_date: 1584748800 |
| length: 10 |
| |
| |
| |
| |
| |
| |