Merge branch 'master' of git.wanmeizhensuo.com:ML/ffm-baseline

change con_sql argument

eda/recommended_indexs/getClickZeroUidRateDetail.py

@@ -21,82 +21,109 @@ my_tm5 = int(my_date5.strftime("%s"))
 
 def get_rate_detail(platform):
 	if platform == "ios":
-		self.platform = "='App Store'"
+		platform = "='App Store'"
 	elif platform == "android":
-		self.platform = "!='App Store'"
+		platform = "!='App Store'"
 	else:
-		self.platform = " is not null"
+		platform = " is not null"
 	db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_test')
 	cursor = db.cursor()
-	sql_tmp = "create temporary table if not exists tmp_table as \
-	(select distinct(device_id) from data_feed_exposure \
+	sql = "select count(distinct(device_id)) from data_feed_exposure \
 	where device_type{0} \
 	and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day) \
 	and device_id not in \
 	(select distinct(device_id) from data_feed_click \
 	where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day) \
-		and device_type{1}))".format(platform,platform.replace(' ','') if platform[-2]=='e' else platform)
-	sql_last = "select count(distinct(device_id)) from tmp_table\
-	where device_id in \
+		and device_type{1}) \
+	and device_id in \
 	(select distinct(device_id) \
 	from data_feed_exposure \
 	where device_id not in \
 	(select distinct(device_id) from data_feed_exposure \
-		where time < {0})) \
+		where time < {2})) \
 	union \
-	select count(distinct(device_id)) from tmp_table\
-	where device_id in \
+	select count(distinct(device_id)) from data_feed_exposure \
+	where device_type{0} \
+	and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day) \
+	and device_id not in \
+	(select distinct(device_id) from data_feed_click \
+	where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day) \
+		and device_type{1}) \
+	and device_id in \
 	(select distinct(device_id) \
 	from data_feed_exposure \
 	where device_id not in \
 	(select distinct(device_id) from data_feed_exposure \
-		where time < {1}) \
+		where time < {3}) \
 	and device_id in \
 	(select distinct(device_id) from data_feed_exposure \
-		where time < {0})) \
+		where time < {2})) \
 	union \
-	select count(distinct(device_id)) from tmp_table\
-	where device_id in \
+	select count(distinct(device_id)) from data_feed_exposure \
+	where device_type{0} \
+	and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day) \
+	and device_id not in \
+	(select distinct(device_id) from data_feed_click \
+	where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day) \
+		and device_type{1}) \
+	and device_id in \
 	(select distinct(device_id) \
 	from data_feed_exposure \
 	where device_id not in \
 	(select distinct(device_id) from data_feed_exposure \
-		where time < {2}) \
+		where time < {4}) \
 	and device_id in \
 	(select distinct(device_id) from data_feed_exposure \
-		where time < {1})) \
+		where time < {3})) \
 	union \
-	select count(distinct(device_id)) from tmp_table\
-	where device_id in \
+	select count(distinct(device_id)) from data_feed_exposure \
+	where device_type{0} \
+	and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day) \
+	and device_id not in \
+	(select distinct(device_id) from data_feed_click \
+	where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day) \
+		and device_type{1}) \
+	and device_id in \
 	(select distinct(device_id) \
 	from data_feed_exposure \
 	where device_id not in \
 	(select distinct(device_id) from data_feed_exposure \
-		where time < {3}) \
+		where time < {5}) \
 	and device_id in \
 	(select distinct(device_id) from data_feed_exposure \
-		where time < {2})) \
+		where time < {4})) \
 	union \
-	select count(distinct(device_id)) from tmp_table\
-	where device_id in \
+	select count(distinct(device_id)) from data_feed_exposure \
+	where device_type{0} \
+	and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day) \
+	and device_id not in \
+	(select distinct(device_id) from data_feed_click \
+	where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day) \
+		and device_type{1}) \
+	and device_id in \
 	(select distinct(device_id) \
 	from data_feed_exposure \
 	where device_id not in \
 	(select distinct(device_id) from data_feed_exposure \
-		where time < {4}) \
+		where time < {6}) \
 	and device_id in \
 	(select distinct(device_id) from data_feed_exposure \
-		where time < {3})) \
+		where time < {5})) \
 	union \
-	select count(distinct(device_id)) from tmp_table\
-	where device_id in \
+	select count(distinct(device_id)) from data_feed_exposure \
+	where device_type{0} \
+	and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day) \
+	and device_id not in \
+	(select distinct(device_id) from data_feed_click \
+	where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day) \
+		and device_type{1}) \
+	and device_id in \
 	(select distinct(device_id) \
 	from data_feed_exposure \
 	where device_id in \
 	(select distinct(device_id) from data_feed_exposure \
-		where time < {4}))".format(my_tm1,my_tm2,my_tm3,my_tm4,my_tm5)
-	cursor.execute(sql_tmp)
-	cursor.execute(sql_last)
+		where time < {6}))".format(platform,platform.replace(' ','') if platform[-2]=='e' else platform,my_tm1,my_tm2,my_tm3,my_tm4,my_tm5)
+	cursor.execute(sql)
 	result = cursor.fetchall()
 	db.close()
 	return result

eda/recommended_indexs/getDiaryImpRate.py

+# -*- coding: UTF-8 -*-
+import pymysql
+
+
+def con_sql(sql):
+	# 从数据库的表里获取数据
+    db = pymysql.connect(host='10.66.157.22', port=4000, user='root', passwd='3SYz54LS9#^9sBvC', db='jerry_test')
+    cursor = db.cursor()
+    cursor.execute(sql)
+    result = cursor.fetchall()
+    db.close()
+    return result
+
+
+#1 获取所有平台的日记曝光占比
+def get_all_diary_imp_rate():
+	sql = "select count(cid) from data_feed_exposure where cid_type='diary' and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day)"
+	diary_imp_count = con_sql(sql)
+	diary_imp_count = diary_imp_count[0][0]
+	sql = "select count(cid) from data_feed_exposure where from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day)"
+	all_imp_count = con_sql(sql)
+	all_imp_count = all_imp_count[0][0]
+	all_diary_imp_rate = diary_imp_count / all_imp_count
+	return ["所有",diary_imp_count,all_imp_count,round(all_diary_imp_rate,4)]
+
+
+#2 获取ios平台的日记曝光占比
+def get_ios_diary_imp_rate():
+	sql = "select count(cid) from data_feed_exposure where cid_type='diary' and device_type='App Store' and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day)"
+	diary_imp_count = con_sql(sql)
+	diary_imp_count = diary_imp_count[0][0]
+	sql = "select count(cid) from data_feed_exposure where device_type='App Store' and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day)"
+	all_imp_count = con_sql(sql)
+	all_imp_count = all_imp_count[0][0]
+	ios_diary_imp_rate = diary_imp_count / all_imp_count
+	return ["苹果",diary_imp_count,all_imp_count,round(ios_diary_imp_rate,4)]
+
+
+#3 获取安卓平台的日记曝光占比
+def get_android_diary_imp_rate():
+	sql = "select count(cid) from data_feed_exposure where cid_type='diary' and device_type!='App Store' and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day)"
+	diary_imp_count = con_sql(sql)
+	diary_imp_count = diary_imp_count[0][0]
+	sql = "select count(cid) from data_feed_exposure where device_type!='App Store' and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -1 day)"
+	all_imp_count = con_sql(sql)
+	all_imp_count = all_imp_count[0][0]
+	android_diary_imp_rate = diary_imp_count / all_imp_count
+	return ["安卓",diary_imp_count,all_imp_count,round(android_diary_imp_rate,4)]
+
+
+
+if __name__ == "__main__":
+	all_diary_imp_rate = get_all_diary_imp_rate()
+	ios_diary_imp_rate = get_ios_diary_imp_rate()
+	android_diary_imp_rate = get_android_diary_imp_rate()

eda/recommended_indexs/getRate.py

 # -*- coding: UTF-8 -*-
 import datetime
 from getAnswerImpRate import get_all_answer_imp_rate,get_ios_answer_imp_rate,get_android_answer_imp_rate
+from getDiaryImpRate import get_all_diary_imp_rate,get_ios_diary_imp_rate,get_android_diary_imp_rate
 from getActivateUidCtr import get_all_click_one_rate,get_ios_click_one_rate,get_android_click_one_rate
 from getClickAnswerUidRate import get_all_click_answer_rate,get_ios_click_answer_rate,get_android_click_answer_rate
 from getClickDiaryUidRate import get_all_click_diary_rate,get_ios_click_diary_rate,get_android_click_diary_rate
@@ -20,10 +21,11 @@ def result2file(fpath):
 内容概览：以下所有数据都是首页的
 1. 比例特征
 	1.1 answer曝光占比(=answer被曝光数/总cid被曝光数)
-	1.2 活跃用户点击率(=有点击用户点击次数/有点击用户曝光次数)
-	1.3 点击answer用户占比(=点击answer用户数/曝光answer用户数)
-	1.4 点击diary用户占比(=点击diary用户数/曝光diary用户数)
-	1.5 无点击用户占比(=无点击用户数/有曝光用户数)
+	1.2 diary曝光占比(=diary被曝光数/总cid被曝光数)
+	1.3 活跃用户点击率(=有点击用户点击次数/有点击用户曝光次数)
+	1.4 点击answer用户占比(=点击answer用户数/曝光answer用户数)
+	1.5 点击diary用户占比(=点击diary用户数/曝光diary用户数)
+	1.6 无点击用户占比(=无点击用户数/有曝光用户数)
 2.Top特征
 	2.1 用户点击次数分布(第一列：用户点击次数；第二列：独立用户数量)
 	2.2 Top 100 diary (sorted by ctr)
@@ -49,7 +51,19 @@ def result2file(fpath):
 		f.write('\n')
 		print("1.1已将answer曝光占比存入文件")
 
-		f.write("#1.2活跃用户点击率(=有点击用户点击次数/有点击用户曝光次数)\n")
+		f.write("#1.2diary曝光占比(=diary被曝光数/总cid被曝光数)\n")
+		f.write(tplt.format("平台","diary被曝光数","总cid被曝光数","diary被曝光数占比"))
+		all_diary_imp_rate = get_all_diary_imp_rate()
+		ios_diary_imp_rate = get_ios_diary_imp_rate()
+		android_diary_imp_rate = get_android_diary_imp_rate()
+		lst = [all_diary_imp_rate,ios_diary_imp_rate,android_diary_imp_rate]
+		for i in lst:
+			line = tplt.format(i[0],i[1],i[2],i[3])
+			f.write(line)
+		f.write('\n')
+		print("1.2已将diary曝光占比存入文件")
+
+		f.write("#1.3活跃用户点击率(=有点击用户点击次数/有点击用户曝光次数)\n")
 		f.write(tplt.format("平台","active用户点击次数","active用户曝光次数","active用户点击率"))
 		all_click_one_rate = get_all_click_one_rate()
 		ios_click_one_rate = get_ios_click_one_rate()
@@ -59,9 +73,9 @@ def result2file(fpath):
 			line = tplt.format(i[0],i[1],i[2],i[3])
 			f.write(line)
 		f.write('\n')
-		print("1.2已将活跃用户点击率存入文件")
+		print("1.3已将活跃用户点击率存入文件")
 
-		f.write("#1.3点击answer用户占比(=点击answer用户数/曝光answer用户数)\n")
+		f.write("#1.4点击answer用户占比(=点击answer用户数/曝光answer用户数)\n")
 		f.write(tplt.format("平台","点击answer用户数","曝光answer用户数","击answer用户占比"))
 		all_click_answer_rate = get_all_click_answer_rate()
 		ios_click_answer_rate = get_ios_click_answer_rate()
@@ -71,9 +85,9 @@ def result2file(fpath):
 			line = tplt.format(i[0],i[1],i[2],i[3])
 			f.write(line)
 		f.write('\n')
-		print("1.3已将点击answer用户占比存入文件")
+		print("1.4已将点击answer用户占比存入文件")
 
-		f.write("#1.4点击diary用户占比(=点击diary用户数/曝光diary用户数)\n")
+		f.write("#1.5点击diary用户占比(=点击diary用户数/曝光diary用户数)\n")
 		f.write(tplt.format("平台","点击diary用户数","曝光diary用户数","击diary用户占比"))
 		all_click_diary_rate = get_all_click_diary_rate()
 		ios_click_diary_rate = get_ios_click_diary_rate()
@@ -83,9 +97,9 @@ def result2file(fpath):
 			line = tplt.format(i[0],i[1],i[2],i[3])
 			f.write(line)
 		f.write('\n')
-		print("1.4已将点击diary用户占比存入文件")
+		print("1.5已将点击diary用户占比存入文件")
 
-		f.write("#1.5无点击用户占比(=无点击用户数/有曝光用户数)\n")
+		f.write("#1.6无点击用户占比(=无点击用户数/有曝光用户数)\n")
 		f.write(tplt.format("平台","no点击用户数","have曝光用户数","no点击用户占比"))
 		all_click_zero_rate = get_all_click_zero_rate()
 		ios_click_zero_rate = get_ios_click_zero_rate()
@@ -95,7 +109,7 @@ def result2file(fpath):
 			line = tplt.format(i[0],i[1],i[2],i[3])
 			f.write(line)
 		f.write('\n')
-		print("1.5已将无点击用户占比存入文件")
+		print("1.6已将无点击用户占比存入文件")
 
 
 def main():