模型上线

spark/featureEngSk.py

@@ -1103,19 +1103,17 @@ if __name__ == '__main__':
     spark = get_spark("SERVICE_FEATURE_CSV_EXPORT_SK")
     spark.sparkContext.setLogLevel("ERROR")
     #获取点击曝光数据
-    # clickDF, expDF, ratingDF, startDay, endDay = get_click_exp_rating_df(trainDays, spark)
+    clickDF, expDF, ratingDF, startDay, endDay = get_click_exp_rating_df(trainDays, spark)
 
     #item Es Feature
-    # itemEsFeatureDF = get_item_es_feature_df()
+    itemEsFeatureDF = get_item_es_feature_df()
 
     #计算 item 统计特征
-    # clickStaticFeatures, expStaticFeatures = getItemStaticFeatures(itemStatisticStartDays + trainDays + 1, startDay, endDay)
+    clickStaticFeatures, expStaticFeatures = getItemStaticFeatures(itemStatisticStartDays + trainDays + 1, startDay, endDay)
 
     #计算线上推理 item 统计特征
     predictClickStaticFeatures, predictExpStaticFeatures = getPredictItemStaticFeatures(itemStatisticStartDays)
 
-    predictClickStaticFeatures.show(100, False)
-    predictExpStaticFeatures.show(100, False)
 
     #user Profile Feature
     userProfileFeatureDF = getUserProfileFeature(spark, addDays(-trainDays - 1, format = "%Y-%m-%d"), addDays(-1, format = "%Y-%m-%d"))
@@ -1221,6 +1219,15 @@ if __name__ == '__main__':
 
     print("训练数据写入 耗时s:{}".format(time.time() - write_time_start))
 
+    #存入线上预测特征
+    # card_id | ITEM_NUMERIC_click_count_sum | ITEM_NUMERIC_click_count_avg | ITEM_NUMERIC_click_count_stddev
+    predictClickStaticDF = predictClickStaticFeatures.toPandas()
+    # card_id | ITEM_NUMERIC_exp_count_sum | ITEM_NUMERIC_exp_count_avg | ITEM_NUMERIC_exp_count_stddev
+    predictExpStaticDF = predictExpStaticFeatures.toPandas()
+    #
+
+
+
     print("总耗时:{} mins".format((time.time() - start)/60))
 
     spark.stop()
\ No newline at end of file