Merge branch 'master' of https://git.yishihui.com/linfan/OffLineRec

calCtr.py

@@ -43,10 +43,11 @@ if __name__=="__main__":
             item_info = data_dict[vid]
             view_users = item_info[0]+view_users
             view_pv = item_info[1]+view_pv
-            play_users = item_info[2]+play_pv
-            share_users = item_info[3]+share_users
-            share_pv = item_info[4]+share_pv
-            return_users = item_info[5]+return_users 
+            play_users = item_info[2]+play_users
+            play_pv = item_info[3]+play_pv
+            share_users = item_info[4]+share_users
+            share_pv = item_info[5]+share_pv
+            return_users = item_info[6]+return_users 
             data_dict[vid] = (view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users)
     #print(data_dict.items())
     info_dict = {}
@@ -61,13 +62,15 @@ if __name__=="__main__":
         share_pv = v[5]
         share_users = v[4]
         play_users = v[2]
+        play_pv = v[3]
         #print("return_users:", return_users) 
         k_score = float(return_users)/(float(view_users)+10)
+        k_score2 = float(return_users)/(float(view_pv)+10)
         #print(k_score)
         share_score = float(share_pv)/(float(view_pv)+5)
         backrate = float(return_users)/(float(view_users)+5)
         #print(k, k_score, share_score*backrate, share_score, backrate) 
-        score_info = [share_score, share_score*backrate, share_score, backrate]
+        score_info = [k_score2, share_score*backrate, share_score, backrate, view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users]
         k = "k_p:"+k
         score_info = json.dumps(score_info)
         info_dict[k] = score_info

calCtr1days.py

@@ -43,10 +43,11 @@ if __name__=="__main__":
             item_info = data_dict[vid]
             view_users = item_info[0]+view_users
             view_pv = item_info[1]+view_pv
-            play_users = item_info[2]+play_pv
-            share_users = item_info[3]+share_users
-            share_pv = item_info[4]+share_pv
-            return_users = item_info[5]+return_users 
+            play_users = item_info[2]+play_users
+            play_pv = item_info[3]+play_pv
+            share_users = item_info[4]+share_users
+            share_pv = item_info[5]+share_pv
+            return_users = item_info[6]+return_users 
             data_dict[vid] = (view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users)
     #print(data_dict.items())
     info_dict = {}
@@ -63,12 +64,13 @@ if __name__=="__main__":
         play_users = v[2]
         #print("return_users:", return_users) 
         k_score = float(return_users)/(float(view_users)+10)
+        k_score2 = float(return_users)/(float(view_pv)+10)
         #print(k_score)
         share_score = float(share_pv)/(float(view_pv)+50)
         backrate = float(return_users)/(float(view_users)+10)
         ctr_score = float(play_pv)/(float(view_pv)+50)
         #print(k, k_score, share_score*backrate, share_score, backrate) 
-        score_info = [k_score, share_score*backrate, share_score, backrate, ctr_score, view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users]
+        score_info = [k_score2, share_score*backrate, share_score, backrate, ctr_score, view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users]
         #k = "k_p3:"+k
         score_info = json.dumps(score_info)
         info_dict[k] = score_info

calCtr3days.py

@@ -43,10 +43,11 @@ if __name__=="__main__":
             item_info = data_dict[vid]
             view_users = item_info[0]+view_users
             view_pv = item_info[1]+view_pv
-            play_users = item_info[2]+play_pv
-            share_users = item_info[3]+share_users
-            share_pv = item_info[4]+share_pv
-            return_users = item_info[5]+return_users 
+            play_users = item_info[2]+play_users
+            play_pv =  item_info[3]+play_pv
+            share_users = item_info[4]+share_users
+            share_pv = item_info[5]+share_pv
+            return_users = item_info[6]+return_users 
             data_dict[vid] = (view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users)
     #print(data_dict.items())
     info_dict = {}
@@ -63,12 +64,13 @@ if __name__=="__main__":
         play_users = v[2]
         #print("return_users:", return_users) 
         k_score = float(return_users)/(float(view_users)+30)
+        k_score2 = float(return_users)/(float(view_pv)+30)
         #print(k_score)
         share_score = float(share_pv)/(float(view_pv)+100)
         backrate = float(return_users)/(float(view_users)+30)
         ctr_score = float(play_pv)/(float(view_pv)+100)
         #print(k, k_score, share_score*backrate, share_score, backrate) 
-        score_info = [k_score, share_score*backrate, share_score, backrate, ctr_score, view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users]
+        score_info = [k_score2, share_score*backrate, share_score, backrate, ctr_score, view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users]
         #k = "k_p3:"+k
         score_info = json.dumps(score_info)
         info_dict[k] = score_info

calHourCtr.py

@@ -44,10 +44,11 @@ if __name__=="__main__":
             item_info = data_dict[vid]
             view_users = item_info[0]+view_users
             view_pv = item_info[1]+view_pv
-            play_users = item_info[2]+play_pv
-            share_users = item_info[3]+share_users
-            share_pv = item_info[4]+share_pv
-            return_users = item_info[5]+return_users 
+            play_users = item_info[2]+play_users
+            play_pv = item_info[3]+play_pv
+            share_users = item_info[4]+share_users
+            share_pv = item_info[5]+share_pv
+            return_users = item_info[6]+return_users 
             data_dict[vid] = (view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users)
     #print(data_dict.items())
     f.close()
@@ -63,14 +64,15 @@ if __name__=="__main__":
         share_pv = v[5]
         share_users = v[4]
         play_users = v[2]
+        play_pv = v[3]
         #print("return_users:", return_users) 
         k_score = float(return_users)/(float(view_users)+5)
-        #k_score2 = float(return_users)/(float(share_pv)+5)
+        k_score2 = float(return_users)/(float(view_pv)+5)
         #print(k_score)
         share_score = float(share_pv)/(float(view_pv)+5)
         backrate = float(return_users)/(float(view_users)+5)
         #print(k, k_score, share_score*backrate, share_score, backrate) 
-        score_info = [share_score, share_score*backrate, share_score, backrate]
+        score_info = [k_score2, share_score*backrate, share_score, backrate, view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users]
         score_info = json.dumps(score_info)
         info_dict[k] = score_info
         f.write(k+"\t"+score_info+"\n")

calHourData.py

@@ -44,10 +44,11 @@ if __name__=="__main__":
             item_info = data_dict[vid]
             view_users = item_info[0]+view_users
             view_pv = item_info[1]+view_pv
-            play_users = item_info[2]+play_pv
-            share_users = item_info[3]+share_users
-            share_pv = item_info[4]+share_pv
-            return_users = item_info[5]+return_users 
+            play_users = item_info[2]+play_users
+            play_pv =  item_info[3]+play_pv
+            share_users = item_info[4]+share_users
+            share_pv = item_info[5]+share_pv
+            return_users = item_info[6]+return_users 
             data_dict[vid] = (view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users)
     #print(data_dict.items())
     f.close()
@@ -65,12 +66,12 @@ if __name__=="__main__":
         play_users = v[2]
         #print("return_users:", return_users) 
         k_score = float(return_users)/(float(view_users)+5)
-        #k_score2 = float(return_users)/(float(share_pv)+5)
+        k_score2 = float(return_users)/(float(view_pv)+5)
         #print(k_score)
         share_score = float(share_pv)/(float(view_pv)+5)
         backrate = float(return_users)/(float(view_users)+5)
         #print(k, k_score, share_score*backrate, share_score, backrate) 
-        score_info = [k_score, share_score*backrate, share_score, backrate, view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users]
+        score_info = [k_score2, share_score*backrate, share_score, backrate, view_users, view_pv, play_users, play_pv, share_users, share_pv, return_users]
         score_info = json.dumps(score_info)
         info_dict[k] = score_info
         f.write(k+"\t"+score_info+"\n")

clean.sh

@@ -2,7 +2,7 @@
 source ~/.bash_profile
 source ~/.bashrc
 
-last3day=`date  +"%Y%m%d" -d -3days`
+last3day=`date  +"%Y%m%d" -d -4days`
 
 merge_path=./data/merge_score_${last3day}'*'
 video_data_path=./data/video_data_${last3day}'*'

compose_score_3day.py

@@ -86,12 +86,12 @@ if __name__=="__main__":
         kid2 = "kp_3:"+k
         kid3 = "kp_4:"+k
         f3.write(kid2+"\t"+total_item_info+"\n")
-        res_dict[kid2] = total_item_info
+        #res_dict[kid2] = total_item_info
         res_dict2[kid3] = total_item_info
-    if len(res_dict)>0:
-        redis_helper = RedisHelper()
-        redis_helper.update_batch_setnx_key(res_dict, 60*60*24*15)
+    #if len(res_dict)>0:
+    #    redis_helper = RedisHelper()
+    #    redis_helper.update_batch_setnx_key(res_dict, 60*60*24*15)
     if len(res_dict2)>0:
         redis_helper = RedisHelper()
-        redis_helper.update_batch_setnx_key(res_dict2, 60*60*24*15)
+        redis_helper.update_batch_setnx_key(res_dict2, 60*60*24*3)
     f2.close()

run_ctr.sh

@@ -47,7 +47,7 @@ then
     echo "[ERROR] echo 'calCtr.py"
     exit 255
 fi
-python compose_score.py ${nowday}
+python compose_score2.py ${nowday}
 if [ $? -ne 0 ];
 then
     msg = "[ERROR] cal compose_score "