Commit 48337495 authored by 郭羽's avatar 郭羽

搜索指标统计

parent da47d739
......@@ -2,7 +2,7 @@ source /srv/envs/esmm/bin/activate
/opt/spark/bin/spark-submit --master yarn --deploy-mode client --queue root.strategy --driver-memory 1g --executor-memory 2g --executor-cores 1 --num-executors 2 --conf spark.default.parallelism=50 --conf spark.storage.memoryFraction=0.5 --conf spark.shuffle.memoryFraction=0.3 --conf spark.locality.wait=0 --jars /srv/apps/spark-connector_2.11-1.9.0-rc2.jar,/srv/apps/mysql-connector-java-5.1.38.jar /srv/apps/meta_base_code/task/conent_detail_page_grayscale_ctr.py
/opt/spark/bin/spark-submit --master yarn --deploy-mode client --queue root.strategy --driver-memory 1g --executor-memory 2g --executor-cores 1 --num-executors 2 --conf spark.default.parallelism=50 --conf spark.storage.memoryFraction=0.5 --conf spark.shuffle.memoryFraction=0.3 --conf spark.locality.wait=0 --jars /srv/apps/spark-connector_2.11-1.9.0-rc2.jar,/srv/apps/mysql-connector-java-5.1.38.jar /srv/apps/meta_base_code/task/recommend_strategy_d.py
/opt/spark/bin/spark-submit --master yarn --deploy-mode client --queue root.strategy --driver-memory 1g --executor-memory 2g --executor-cores 1 --num-executors 2 --conf spark.default.parallelism=50 --conf spark.storage.memoryFraction=0.5 --conf spark.shuffle.memoryFraction=0.3 --conf spark.locality.wait=0 --jars /srv/apps/spark-connector_2.11-1.9.0-rc2.jar,/srv/apps/mysql-connector-java-5.1.38.jar /srv/apps/meta_base_code/task/recommend_strategy_fix.py
#/opt/spark/bin/spark-submit --master yarn --deploy-mode client --queue root.strategy --driver-memory 1g --executor-memory 2g --executor-cores 1 --num-executors 2 --conf spark.default.parallelism=50 --conf spark.storage.memoryFraction=0.5 --conf spark.shuffle.memoryFraction=0.3 --conf spark.locality.wait=0 --jars /srv/apps/spark-connector_2.11-1.9.0-rc2.jar,/srv/apps/mysql-connector-java-5.1.38.jar /srv/apps/meta_base_code/task/search_strategy_d.py
/opt/spark/bin/spark-submit --master yarn --deploy-mode client --queue root.strategy --driver-memory 1g --executor-memory 2g --executor-cores 1 --num-executors 2 --conf spark.default.parallelism=50 --conf spark.storage.memoryFraction=0.5 --conf spark.shuffle.memoryFraction=0.3 --conf spark.locality.wait=0 --jars /srv/apps/spark-connector_2.11-1.9.0-rc2.jar,/srv/apps/mysql-connector-java-5.1.38.jar /srv/apps/meta_base_code/task/search_strategy_d.py
......@@ -128,6 +128,8 @@ for t in range(0, task_days):
AND partition_date < '{end_date}'
AND action = 'on_click_card'
AND params['page_name'] = 'search_home'
AND params['in_page_pos'] <> '更美热门榜'
union all
SELECT cl_id,
......
......@@ -80,61 +80,61 @@ for t in range(0, task_days):
today_str = now.strftime("%Y%m%d")
yesterday_str = (now + datetime.timedelta(days=-1)).strftime("%Y%m%d")
one_week_age_str = (now + datetime.timedelta(days=-7)).strftime("%Y%m%d")
sql_dev_device_id = """
SELECT partition_date,device_id
FROM
(--找出user_id当天活跃的第一个设备id
SELECT user_id,partition_date,
if(size(device_list) > 0, device_list [ 0 ], '') AS device_id
FROM online.ml_user_updates
WHERE partition_date>='{yesterday_str}' AND partition_date<'{today_str}'
)t1
JOIN
( --医生账号
SELECT distinct user_id
FROM online.tl_hdfs_doctor_view
WHERE partition_date = '{yesterday_str}'
--马甲账号/模特用户
UNION ALL
SELECT user_id
FROM ml.ml_c_ct_ui_user_dimen_d
WHERE partition_day = '{yesterday_str}'
AND (is_puppet = 'true' or is_classifyuser = 'true')
UNION ALL
--公司内网覆盖用户
select distinct user_id
from dim.dim_device_user_staff
UNION ALL
--登陆过医生设备
SELECT distinct t1.user_id
FROM
(
SELECT user_id, v.device_id as device_id
FROM online.ml_user_history_detail
LATERAL VIEW EXPLODE(device_history_list) v AS device_id
WHERE partition_date = '{yesterday_str}'
)t1
JOIN
(
SELECT device_id
FROM online.ml_device_history_detail
WHERE partition_date = '{yesterday_str}'
AND is_login_doctor = '1'
)t2
ON t1.device_id = t2.device_id
)t2
on t1.user_id=t2.user_id
group by partition_date,device_id
""".format(yesterday_str=yesterday_str, today_str=today_str)
print(sql_dev_device_id)
dev_df = spark.sql(sql_dev_device_id)
dev_df_view = dev_df.createOrReplaceTempView("dev_view")
dev_df.cache()
dev_df.show(1)
sql_res = dev_df.collect()
# sql_dev_device_id = """
# SELECT partition_date,device_id
# FROM
# (--找出user_id当天活跃的第一个设备id
# SELECT user_id,partition_date,
# if(size(device_list) > 0, device_list [ 0 ], '') AS device_id
# FROM online.ml_user_updates
# WHERE partition_date>='{yesterday_str}' AND partition_date<'{today_str}'
# )t1
# JOIN
# ( --医生账号
# SELECT distinct user_id
# FROM online.tl_hdfs_doctor_view
# WHERE partition_date = '{yesterday_str}'
#
# --马甲账号/模特用户
# UNION ALL
# SELECT user_id
# FROM ml.ml_c_ct_ui_user_dimen_d
# WHERE partition_day = '{yesterday_str}'
# AND (is_puppet = 'true' or is_classifyuser = 'true')
#
# UNION ALL
# --公司内网覆盖用户
# select distinct user_id
# from dim.dim_device_user_staff
#
# UNION ALL
# --登陆过医生设备
# SELECT distinct t1.user_id
# FROM
# (
# SELECT user_id, v.device_id as device_id
# FROM online.ml_user_history_detail
# LATERAL VIEW EXPLODE(device_history_list) v AS device_id
# WHERE partition_date = '{yesterday_str}'
# )t1
# JOIN
# (
# SELECT device_id
# FROM online.ml_device_history_detail
# WHERE partition_date = '{yesterday_str}'
# AND is_login_doctor = '1'
# )t2
# ON t1.device_id = t2.device_id
# )t2
# on t1.user_id=t2.user_id
# group by partition_date,device_id
# """.format(yesterday_str=yesterday_str, today_str=today_str)
# print(sql_dev_device_id)
# dev_df = spark.sql(sql_dev_device_id)
# dev_df_view = dev_df.createOrReplaceTempView("dev_view")
# dev_df.cache()
# dev_df.show(1)
# sql_res = dev_df.collect()
# for res in sql_res:
# print(res)
......@@ -266,10 +266,10 @@ for t in range(0, task_days):
LEFT JOIN spam_pv
on spam_pv.device_id=t1.cl_id
LEFT JOIN dev_view
on t1.partition_date=dev_view.partition_date and t1.cl_id=dev_view.device_id
--LEFT JOIN dev_view
--on t1.partition_date=dev_view.partition_date and t1.cl_id=dev_view.device_id
WHERE (spam_pv.device_id IS NULL or spam_pv.device_id ='')
and (dev_view.device_id is null or dev_view.device_id ='')
--and (dev_view.device_id is null or dev_view.device_id ='')
GROUP BY t1.partition_date,t2.active_type,device_os_type,channel
)t
)t3
......@@ -361,10 +361,10 @@ for t in range(0, task_days):
on t1.cl_id=dev.device_id and t1.partition_date = dev.partition_date
LEFT JOIN spam_pv
on spam_pv.device_id=t1.cl_id
LEFT JOIN dev_view
on t1.partition_date=dev_view.partition_date and t1.cl_id=dev_view.device_id
--LEFT JOIN dev_view
--on t1.partition_date=dev_view.partition_date and t1.cl_id=dev_view.device_id
WHERE (spam_pv.device_id IS NULL or spam_pv.device_id ='')
and (dev_view.device_id is null or dev_view.device_id ='')
--and (dev_view.device_id is null or dev_view.device_id ='')
GROUP BY t1.partition_date,active_type,device_os_type,channel
)t4
on t3.partition_date=t4.partition_date and t3.active_type=t4.active_type and t3.device_os_type = t4.device_os_type AND t3.channel = t4.channel
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment