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

增加日记队列比较功能

eda/recommended_indexs/getClickZeroUidDetail.py

@@ -31,10 +31,10 @@ def get_click_zero_uid_count(platform):
 	cursor = db.cursor()
 	sql = "select '0-7' as label,count(distinct(device_id)) from data_feed_exposure2 \
 	where device_type{0} \
-	and stat_date = get_yesterday_date() \
+	and stat_date = '{7}' \
 	and device_id not in \
 	(select distinct(device_id) from data_feed_click2 \
-	where stat_date = get_yesterday_date() \
+	where stat_date = '{7}' \
 		and device_type{1}) \
 	and device_id in \
 	(select distinct(device_id) \
@@ -45,10 +45,10 @@ def get_click_zero_uid_count(platform):
 	union all \
 	select '7-14' as label,count(distinct(device_id)) from data_feed_exposure2 \
 	where device_type{0} \
-	and stat_date = get_yesterday_date() \
+	and stat_date = '{7}' \
 	and device_id not in \
 	(select distinct(device_id) from data_feed_click2 \
-	where stat_date = get_yesterday_date() \
+	where stat_date = '{7}' \
 		and device_type{1}) \
 	and device_id in \
 	(select distinct(device_id) \
@@ -62,10 +62,10 @@ def get_click_zero_uid_count(platform):
 	union all \
 	select '14-30' as label,count(distinct(device_id)) from data_feed_exposure2 \
 	where device_type{0} \
-	and stat_date = get_yesterday_date() \
+	and stat_date = '{7}' \
 	and device_id not in \
 	(select distinct(device_id) from data_feed_click2 \
-	where stat_date = get_yesterday_date() \
+	where stat_date = '{7}' \
 		and device_type{1}) \
 	and device_id in \
 	(select distinct(device_id) \
@@ -79,10 +79,10 @@ def get_click_zero_uid_count(platform):
 	union all \
 	select '30-60' as label,count(distinct(device_id)) from data_feed_exposure2 \
 	where device_type{0} \
-	and stat_date = get_yesterday_date() \
+	and stat_date = '{7}' \
 	and device_id not in \
 	(select distinct(device_id) from data_feed_click2 \
-	where stat_date = get_yesterday_date() \
+	where stat_date = '{7}' \
 		and device_type{1}) \
 	and device_id in \
 	(select distinct(device_id) \
@@ -96,10 +96,10 @@ def get_click_zero_uid_count(platform):
 	union all \
 	select '60-90' as label,count(distinct(device_id)) from data_feed_exposure2 \
 	where device_type{0} \
-	and stat_date = get_yesterday_date() \
+	and stat_date = '{7}' \
 	and device_id not in \
 	(select distinct(device_id) from data_feed_click2 \
-	where stat_date = get_yesterday_date() \
+	where stat_date = '{7}' \
 		and device_type{1}) \
 	and device_id in \
 	(select distinct(device_id) \
@@ -113,17 +113,17 @@ def get_click_zero_uid_count(platform):
 	union all \
 	select '90+' as label,count(distinct(device_id)) from data_feed_exposure2 \
 	where device_type{0} \
-	and stat_date = get_yesterday_date() \
+	and stat_date = '{7}' \
 	and device_id not in \
 	(select distinct(device_id) from data_feed_click2 \
-	where stat_date = get_yesterday_date() \
+	where stat_date = '{7}' \
 		and device_type{1}) \
 	and device_id in \
 	(select distinct(device_id) \
 	from data_feed_exposure2 \
 	where device_id in \
 	(select distinct(device_id) from data_feed_exposure2 \
-		where stat_date < {6}))".format(platform,platform.replace(' ','') if platform[-2]=='e' else platform,my_date1,my_date2,my_date3,my_date4,my_date5)
+		where stat_date < {6}))".format(platform,platform.replace(' ','') if platform[-2]=='e' else platform,my_date1,my_date2,my_date3,my_date4,my_date5,get_yesterday_date())
 	cursor.execute(sql)
 	result = cursor.fetchall()
 	db.close()

eda/recommended_indexs/main.py

@@ -73,18 +73,18 @@ click_everything_result = [click_everything_all,click_everything_ios,click_every
 print("已获取有点击用户占比")
 
 #1.9 无点击用户数分布(=无点击用户∩激活用户 / 激活用户数)   #注意：(]里面的数字指的是距离当前时间的天数
-try:
-	click_zero_uid_detail_all = get_click_zero_uid_rate_detail("all")
-	click_zero_uid_detail_all["platform"] = "所有"
-	click_zero_uid_detail_ios = get_click_zero_uid_rate_detail("ios")
-	click_zero_uid_detail_ios["platform"] = "苹果"
-	click_zero_uid_detail_android = get_click_zero_uid_rate_detail("android")
-	click_zero_uid_detail_android["platform"] = "安卓"
-	click_zero_uid_detail_result = [click_zero_uid_detail_all,click_zero_uid_detail_ios,click_zero_uid_detail_android]
-	print("已获取无点击用户数激活日期分布")
-except:
-	click_zero_uid_detail_result = []
-	print("GC life time is shorter than transaction duration")
+# try:
+# 	click_zero_uid_detail_all = get_click_zero_uid_rate_detail("all")
+# 	click_zero_uid_detail_all["platform"] = "所有"
+# 	click_zero_uid_detail_ios = get_click_zero_uid_rate_detail("ios")
+# 	click_zero_uid_detail_ios["platform"] = "苹果"
+# 	click_zero_uid_detail_android = get_click_zero_uid_rate_detail("android")
+# 	click_zero_uid_detail_android["platform"] = "安卓"
+# 	click_zero_uid_detail_result = [click_zero_uid_detail_all,click_zero_uid_detail_ios,click_zero_uid_detail_android]
+# 	print("已获取无点击用户数激活日期分布")
+# except:
+# 	click_zero_uid_detail_result = []
+# 	print("GC life time is shorter than transaction duration")
 
 
 #==========================================================================================
@@ -138,7 +138,6 @@ def result2file():
 	1.6 点击question用户占比(=点击question用户数/曝光question用户数)
 	1.7 点击diary用户占比(=点击diary用户数/曝光diary用户数)
 	1.8 有点击用户占比(=有点击用户数/有曝光用户数)
-	1.9 无点击用户数分布(=无点击用户∩激活用户 / 激活用户数)   #注意：平台一排的数字指的是距离当前时间的天数
 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)
@@ -205,20 +204,20 @@ def result2file():
 		for i in click_everything_result:
 			line = tplt.format(i[0],i[1],i[2],"{}%".format(round(i[3]*100,2)))
 			f.write(line)
-		f.write('\n')
-		if click_zero_uid_detail_result != []:
-			f.write("#1.9无点击用户数分布(=无点击用户∩激活用户 / 激活用户数)   #注意：(]里面的数字指的是距离当前时间的天数\n")
-			f.write("平台"+'\t\t'+"0-7"+'\t\t'+"7-14"+'\t\t'+ \
-				"14-30"+'\t\t'+"30-60"+'\t\t'+"60-90"+'\t\t'+"90+"+'\n')
-			for i in click_zero_uid_detail_result:
-				f.write(i["platform"]+'\t\t'+\
-					"{}%".format(round(i["0-7"]*100,2))+'\t\t'+\
-					"{}%".format(round(i["7-14"]*100,2))+'\t\t'+\
-					"{}%".format(round(i["14-30"]*100,2))+'\t\t'+\
-					"{}%".format(round(i["30-60"]*100,2))+'\t\t'+\
-					"{}%".format(round(i["60-90"]*100,2))+'\t\t'+\
-					"{}%".format(round(i["90+"]*100,2))+'\n')
-			f.write('\n\n\n')
+		# f.write('\n')
+		# if click_zero_uid_detail_result != []:
+		# 	f.write("#1.9无点击用户数分布(=无点击用户∩激活用户 / 激活用户数)   #注意：(]里面的数字指的是距离当前时间的天数\n")
+		# 	f.write("平台"+'\t\t'+"0-7"+'\t\t'+"7-14"+'\t\t'+ \
+		# 		"14-30"+'\t\t'+"30-60"+'\t\t'+"60-90"+'\t\t'+"90+"+'\n')
+		# 	for i in click_zero_uid_detail_result:
+		# 		f.write(i["platform"]+'\t\t'+\
+		# 			"{}%".format(round(i["0-7"]*100,2))+'\t\t'+\
+		# 			"{}%".format(round(i["7-14"]*100,2))+'\t\t'+\
+		# 			"{}%".format(round(i["14-30"]*100,2))+'\t\t'+\
+		# 			"{}%".format(round(i["30-60"]*100,2))+'\t\t'+\
+		# 			"{}%".format(round(i["60-90"]*100,2))+'\t\t'+\
+		# 			"{}%".format(round(i["90+"]*100,2))+'\n')
+		f.write('\n\n\n')
 		
 #==========================================================================================
 		tplt = "{0:^10}\t{1:^10}\n"