模型上线

spark/featureEngSk.py

@@ -51,6 +51,13 @@ def getRedisConn():
     # conn = redis.Redis(host="172.18.51.10", port=6379,db=0) #test
     return conn
 
+def getRedisConn1():
+    pool = redis.ConnectionPool(host="172.16.40.133",password="ReDis!GmTx*0aN6",port=6379,db=0)
+    conn = redis.Redis(connection_pool=pool)
+    # conn = redis.Redis(host="172.16.50.145", port=6379, password="XfkMCCdWDIU%ls$h",db=0)
+    # conn = redis.Redis(host="172.18.51.10", port=6379,db=0) #test
+    return conn
+
 def parseTags(tags,i):
     tags_arr = tags.split(",")
     if len(tags_arr) >= i:
@@ -1038,7 +1045,7 @@ def get_and_save_card_feature(itemEsFeatureDF, predictClickStaticFeatures, predi
     itemFeature = itemFeature.na.fill(fields_na_value_dict)
 
     itemFeatureDF = itemFeature.toPandas()
-    conn = getRedisConn()
+    conn = getRedisConn1()
     BATCH = 5000
     Key_TMP = 'strategy:model:rank:widedeep:service:feature:tmp'
     def concat_service_feature(row):
@@ -1103,7 +1110,7 @@ def get_and_save_device_feature(spark, fields_na_value_dict, days = 180):
     device_feature_df = device_feature_df.na.fill({'os': '-1'})
     device_feature_df.printSchema()
     device_feature_df = device_feature_df.toPandas()
-    conn = getRedisConn()
+    conn = getRedisConn1()
     BATCH = 5000
     Key_TMP = 'strategy:model:rank:widedeep:device:feature:tmp'
     conn.delete(Key_TMP)
@@ -1180,10 +1187,6 @@ if __name__ == '__main__':
     #计算 item 统计特征
     clickStaticFeatures, expStaticFeatures = getItemStaticFeatures(itemStatisticStartDays + trainDays + 1, startDay, endDay)
 
-    #计算线上推理 item 统计特征
-    predictClickStaticFeatures, predictExpStaticFeatures = getPredictItemStaticFeatures(itemStatisticStartDays)
-
-
     #user Profile Feature
     userProfileFeatureDF = getUserProfileFeature(spark, addDays(-trainDays - 1, format = "%Y-%m-%d"), addDays(-1, format = "%Y-%m-%d"))
 
@@ -1341,7 +1344,10 @@ if __name__ == '__main__':
     #存device_id -> USER_CATEGORY_os,
     get_and_save_device_feature(spark, fields_na_value_dict)
 
-    #存card_id -> ITEM*
+    #计算线上推理 item 统计特征
+    predictClickStaticFeatures, predictExpStaticFeatures = getPredictItemStaticFeatures(itemStatisticStartDays)
+
+    # 存card_id -> ITEM*
     get_and_save_card_feature(itemEsFeatureDF, predictClickStaticFeatures, predictExpStaticFeatures, fields_na_value_dict)
 
     print("总耗时:{} mins".format((time.time() - start)/60))