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

eda/recommended_indexs/func.py

@@ -61,7 +61,7 @@ def get_activate_uid_imp_times(city,ndays=1):
 	and from_unixtime(time,'%Y-%m-%d')=date_add(curdate(), interval -{0} day) \
 	and city_id{1}".format(ndays,city)
 	sql_imp_times = con_sql(sql_imp)[0][0]
-	if city == "beijing":
+	if city == "='beijing'":
 		city = "北京"
 	else:
 		city = "所有"

eda/recommended_indexs/main.py

@@ -57,14 +57,17 @@ click_everything_result = [click_everything_all,click_everything_ios,click_every
 print("已获取有点击用户占比")
 
 #1.8 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
-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")
+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:
+	print("GC life time is shorter than transaction duration")
 
 
 #==========================================================================================
@@ -111,7 +114,7 @@ def result2file():
 	1.5 点击answer用户占比(=点击answer用户数/曝光answer用户数)
 	1.6 点击diary用户占比(=点击diary用户数/曝光diary用户数)
 	1.7 有点击用户占比(=有点击用户数/有曝光用户数)
-	1.8 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
+	1.8 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)
@@ -167,18 +170,20 @@ def result2file():
 			line = tplt.format(i[0],i[1],i[2],"{}%".format(round(i[3]*100,2)))
 			f.write(line)
 		f.write('\n')
-		f.write("#1.8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')
+		if "click_zero_uid_detail_result" in dir():
+			f.write("#1.8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"
 		f.write("#2. Top特征\n")