在Spark 2.2.0和Scala中对数据框的列求和

Question

I am trying to sum columns in the following data frame in Spark/Scala, which was itself created through another data frame. I was using this answer as a guide: How to sum the values of one column of a dataframe in spark/scala

Here's my data, created from another aggregate function and assigned to a data frame:

+-------------+----+----+
|activityLabel| 1_3|4_12|
+-------------+----+----+
|           12|1075|   0|
|            1|   0|3072|
|            6|3072|   0|
|            3|   0|3072|
|            5|3072|   0|
|            9|3072|   0|
|            4|3072|   0|
|            8|3379|   0|
|            7|3072|   0|
|           10|3072|   0|
|           11|3072|   0|
|            2|   0|3072|
+-------------+----+----+

And here's my code to create the dataframe:

def createRangeActivityLabels(df: DataFrame): Unit = {

  val activityRange: List[(Int, Int)] = List((1, 3), (4, 12))

  val exprs: List[Column] = activityRange.map {
    case (x, y) => {
      val newLabel = s"${x}_${y}"
      sum(when($"activityLabel".between(x, y), 0).otherwise(1)).alias(newLabel)
    }
  }

  val df3: DataFrame = df.groupBy($"activityLabel").agg(exprs.head, exprs.tail: _*)
  df3.show

And here's the code to get the sum. All I want to do is sum the columns labelled as 1_3 (exprs.head) and 4_12 (exprs(1))

  val indexedLabel0: Int = df3.agg(sum(exprs.head)).first.getAs[Int](0)
}

I get the following error: org.apache.spark.sql.AnalysisException: It is not allowed to use an aggregate function in the argument of another aggregate function. Please use the inner aggregate function in a sub-query.;;

I have tried multiple solutions to fix this but nothing seems to work. All ideas appreciated. Thanks!

Answer 1

The problem is that exprs.head evaluate to sum(when($"activityLabel".between(x, y), 0).otherwise(1)).alias(newLabel) . And when you try to sum(exprs.head), it will evaluate sum of sum.

I think you only need column name.

val columnsName: List[Column] = activityRange.map {
    case (x, y) => $"${x}_${y}"
}
val indexedLabel0 = df3.agg(sum(columnsName.head)).first.getAs[Long](0)

@user8371915 Thanks to correct me about return type