diff

src/main/scala/com/aliyun/odps/spark/examples/makedata_ad/v20240718/diff_data_20250319.scala

@@ -240,25 +240,16 @@ object diff_data_20250319 {
     // 处理结果
     val featureDiffSum = mutable.Map[String, Double]()
     val featureCount = mutable.Map[String, Int]()
-    val countJsonObject1 = new JSONObject()
-    val countJsonObject2 = new JSONObject()
 
     result.foreach { case (key, (diffSum, count)) =>
       featureDiffSum(key) = diffSum
       featureCount(key) = count
-      countJsonObject1.put(key, diffSum)
-      countJsonObject2.put(key, count)
     }
 
     // 打印结果
     println("featureDiffSum: " + featureDiffSum)
     println("featureCount: " + featureCount)
-    println("countJsonObject1: " + countJsonObject1)
-    println("countJsonObject2: " + countJsonObject2)
 
-    println("json")
-    println(countJsonObject1)
-    println(countJsonObject2)
 
 
     // 输出每个特征的平均差异
@@ -270,12 +261,12 @@ object diff_data_20250319 {
       println(s"  Feature: $feature, Average Diff: $averageDiff")
     }
 
-    val count = joinedRDD.count()
-    println(s"对比总数: $count")
+    val count2 = joinedRDD.count()
+    println(s"对比总数: $count2")
     println("每个特征的差异率:")
     println(featureCount.size)
     featureCount.foreach { case (feature, sum) =>
-      val rateDiff = sum / count
+      val rateDiff = sum / count2
       println(s"  Feature: $feature, Rate Diff: $rateDiff")
     }
   }