nevi_me nevi_me - 17 days ago 15
Scala Question

Spark UserDefinedAggregateFunction: scala.MatchError 0.0 (of class java.lang.Double)

I'm trying to use the

UserDefinedAggregateFunction
on Spark 2.0.2 with Scala, but I'm coming across a match error. I've created the below as a test case, the code that I'm writing does something similar to the below.

I am trying to accumulate a value through an aggregation window. It's not just a cumulative sum, but I need to calculate the number to keep, based on some conditions.

As a test case, I have created an amortisation table, where I have to calculate the opening and closing balance of each month.

The data looks like:

+------+--------+------------+---------+
|Period| Capital|InterestRate|Repayment|
+------+--------+------------+---------+
|201601| 0.00 | 0.10 | 0.00 |
|201602|1000.00 | 0.00 | 0.00 |
|201603|2000.00 | 0.10 | 0.00 |
|201604| 0.00 | 0.10 | -200.00 |
|201605| 0.00 | 0.10 | -200.00 |
|201606| 0.00 | 0.10 | -200.00 |
|201607| 0.00 | 0.10 | -200.00 |
|201608| 0.00 | 0.00 | -200.00 |
|201609| 0.00 | 0.10 | -200.00 |
|201610| 0.00 | 0.10 | -200.00 |
|201611| 0.00 | 0.10 | -200.00 |
|201612| 0.00 | 0.10 | -200.00 |
+------+--------+------------+---------+


I couldn't get the CSV to format properly, but I've added it to a gist here: https://gist.github.com/nevi-me/8b2362a5365e73af947fc13bb5836adc.

I'm trying to calculate the
Opening
and
Closing
balances, and then return the
Closing
balance from the aggregation.

Scala



package me.nevi

import org.apache.spark.sql._
import org.apache.spark.sql.expressions.{MutableAggregationBuffer, UserDefinedAggregateFunction, Window}
import org.apache.spark.sql.types.{StructType, DoubleType, DataType}

object AggregationTest {

object amortisedClosingBalance extends UserDefinedAggregateFunction {
override def inputSchema: StructType = new StructType().add("Capital", DoubleType).add("InterestRate", DoubleType).add("Repayment", DoubleType)

override def bufferSchema: StructType = new StructType().add("Opening", DoubleType).add("Closing", DoubleType)

override def dataType: DataType = new StructType().add("Closing", DoubleType)

override def deterministic: Boolean = true

override def initialize(buffer: MutableAggregationBuffer): Unit = {
buffer.update(0, 0.0)
buffer.update(1, 0.0)
}

override def update(buffer: MutableAggregationBuffer, input: Row): Unit = {
if (!input.isNullAt(0)) {
println(buffer.get(0))
println(buffer.get(1))
buffer.update(0, buffer.getDouble(1))
// (opening + capital) * interestrate - repayment
buffer.update(1, (buffer.getDouble(0) + input.getDouble(0)) * input.getDouble(1) + input.getDouble(2))
} else {
// if first record?
buffer.update(0, input.getDouble(0))
buffer.update(1, input.getDouble(0))
}
}

override def merge(buffer1: MutableAggregationBuffer, buffer2: Row): Unit = {
buffer1.update(0, buffer1.getDouble(0))
buffer1.update(1, buffer1.getDouble(1))
}

override def evaluate(buffer: Row): Any = {
buffer.getDouble(1)
}
}

def main(args: Array[String]): Unit = {
System.setProperty("hadoop.home.dir", "C:/spark")
System.setProperty("spark.sql.warehouse.dir", "file:///tmp/spark-warehouse")

val spark: SparkSession = SparkSession.builder()
.master("local[*]")
.appName("Aggregation Test")
.getOrCreate()

import spark.implicits._

val df = spark.read.option("header", true).csv("file:///d:/interest_calc.csv")

df.show()

val windowSpec = Window.orderBy(df.col("Period"))

val calc = df.withColumn("Closing", amortisedClosingBalance($"Capital", $"InterestRate", $"Repayment").over(windowSpec))

calc.show()

}
}


I get the exception:

scala.MatchError: 0.0 (of class java.lang.Double)
at org.apache.spark.sql.catalyst.CatalystTypeConverters$StructConverter.toCatalystImpl(CatalystTypeConverters.scala:256)
at org.apache.spark.sql.catalyst.CatalystTypeConverters$StructConverter.toCatalystImpl(CatalystTypeConverters.scala:251)
at org.apache.spark.sql.catalyst.CatalystTypeConverters$CatalystTypeConverter.toCatalyst(CatalystTypeConverters.scala:103)
at org.apache.spark.sql.catalyst.CatalystTypeConverters$$anonfun$createToCatalystConverter$2.apply(CatalystTypeConverters.scala:403)
at org.apache.spark.sql.execution.aggregate.ScalaUDAF.eval(udaf.scala:440)
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$SpecificMutableProjection.apply(Unknown Source)
at org.apache.spark.sql.execution.AggregateProcessor.evaluate(WindowExec.scala:1029)
at org.apache.spark.sql.execution.UnboundedPrecedingWindowFunctionFrame.write(WindowExec.scala:822)
at org.apache.spark.sql.execution.WindowExec$$anonfun$15$$anon$1.next(WindowExec.scala:398)
at org.apache.spark.sql.execution.WindowExec$$anonfun$15$$anon$1.next(WindowExec.scala:289)
at org.apache.spark.sql.catalyst.expressions.GeneratedClass$GeneratedIterator.processNext(Unknown Source)
at org.apache.spark.sql.execution.BufferedRowIterator.hasNext(BufferedRowIterator.java:43)
at org.apache.spark.sql.execution.WholeStageCodegenExec$$anonfun$8$$anon$1.hasNext(WholeStageCodegenExec.scala:370)
at org.apache.spark.sql.execution.SparkPlan$$anonfun$4.apply(SparkPlan.scala:246)
at org.apache.spark.sql.execution.SparkPlan$$anonfun$4.apply(SparkPlan.scala:240)
at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$24.apply(RDD.scala:803)
at org.apache.spark.rdd.RDD$$anonfun$mapPartitionsInternal$1$$anonfun$apply$24.apply(RDD.scala:803)
at org.apache.spark.rdd.MapPartitionsRDD.compute(MapPartitionsRDD.scala:38)
at org.apache.spark.rdd.RDD.computeOrReadCheckpoint(RDD.scala:319)
at org.apache.spark.rdd.RDD.iterator(RDD.scala:283)
at org.apache.spark.scheduler.ResultTask.runTask(ResultTask.scala:70)
at org.apache.spark.scheduler.Task.run(Task.scala:86)
at org.apache.spark.executor.Executor$TaskRunner.run(Executor.scala:274)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745)


Does anyone know what I am doing wrong? I initially used Spark 2.0.0, and I came across someone else with a similar issue with UDTF's, and the suggestion was to upgrade to 2.0.1, however after upgrading; my issue still persists.

Answer

You get an exception due to incorrect dataType definition. You declare it as:

StructType(StructField(Closing,DoubleType,true))

when in fact you return a scalar. It should be defined as:

override def dataType: DataType = DoubleType

or you should redefine evalute, for example to:

override def evaluate(buffer: Row): Any = {
  Row(buffer.getDouble(1))
}

The latter one will return a nested column:

 |-- Closing: struct (nullable = true)
 |    |-- Closing: double (nullable = true)

so it is probably not what you're looking for.