-
Notifications
You must be signed in to change notification settings - Fork 539
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Several optimisation w.r.t the new KLL sketches
- Loading branch information
1 parent
0436d5a
commit 9c69f00
Showing
14 changed files
with
384 additions
and
32 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
179 changes: 179 additions & 0 deletions
179
src/main/scala/com/amazon/deequ/analyzers/runners/KLLRunner.scala
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,179 @@ | ||
| /** | ||
| * Copyright 2018 Amazon.com, Inc. or its affiliates. All Rights Reserved. | ||
| * | ||
| * Licensed under the Apache License, Version 2.0 (the "License"). You may not | ||
| * use this file except in compliance with the License. A copy of the License | ||
| * is located at | ||
| * | ||
| * http://aws.amazon.com/apache2.0/ | ||
| * | ||
| * or in the "license" file accompanying this file. This file is distributed on | ||
| * an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either | ||
| * express or implied. See the License for the specific language governing | ||
| * permissions and limitations under the License. | ||
| * | ||
| */ | ||
|
|
||
| package com.amazon.deequ.analyzers.runners | ||
|
|
||
| import com.amazon.deequ.analyzers.{Analyzer, KLLParameters, KLLSketch, KLLState, QuantileNonSample, State, StateLoader, StatePersister} | ||
| import com.amazon.deequ.metrics.Metric | ||
| import org.apache.spark.sql.types.{ByteType, DoubleType, FloatType, IntegerType, LongType, ShortType, StructType} | ||
| import org.apache.spark.sql.{DataFrame, Row} | ||
|
|
||
| @SerialVersionUID(1L) | ||
| abstract class UntypedQuantileNonSample(sketchSize: Int, shrinkingFactor: Double) | ||
| extends Serializable { | ||
|
|
||
| var min: Double = Int.MaxValue.toDouble | ||
| var max: Double = Int.MinValue.toDouble | ||
| var sketch: QuantileNonSample[Double] = new QuantileNonSample[Double](sketchSize, shrinkingFactor) | ||
|
|
||
| def itemAsDouble(item: Any): Double | ||
|
|
||
| def updateUntyped(item: Any): Unit = { | ||
| this.min = math.min(this.min, itemAsDouble(item)) | ||
| this.max = math.max(this.max, itemAsDouble(item)) | ||
| sketch.update(itemAsDouble(item)) | ||
| } | ||
|
|
||
| def mergeUntyped(other: UntypedQuantileNonSample): Unit = { | ||
| this.min = math.min(this.min, other.min) | ||
| this.max = math.max(this.max, other.max) | ||
| this.sketch = this.sketch.merge(other.sketch) | ||
| } | ||
|
|
||
| def asKLLState(): KLLState = { | ||
| KLLState(sketch, max, min) | ||
| } | ||
| } | ||
|
|
||
| @SerialVersionUID(1L) | ||
| class LongQuantileNonSample(sketchSize: Int, shrinkingFactor: Double) | ||
| extends UntypedQuantileNonSample(sketchSize, shrinkingFactor) with Serializable { | ||
| override def itemAsDouble(item: Any): Double = item.asInstanceOf[Long].toDouble | ||
| } | ||
|
|
||
| @SerialVersionUID(1L) | ||
| class IntQuantileNonSample(sketchSize: Int, shrinkingFactor: Double) | ||
| extends UntypedQuantileNonSample(sketchSize, shrinkingFactor) with Serializable { | ||
| override def itemAsDouble(item: Any): Double = item.asInstanceOf[Int].toDouble | ||
| } | ||
|
|
||
| @SerialVersionUID(1L) | ||
| class ShortQuantileNonSample(sketchSize: Int, shrinkingFactor: Double) | ||
| extends UntypedQuantileNonSample(sketchSize, shrinkingFactor) with Serializable { | ||
| override def itemAsDouble(item: Any): Double = item.asInstanceOf[Short].toDouble | ||
| } | ||
|
|
||
| @SerialVersionUID(1L) | ||
| class ByteQuantileNonSample(sketchSize: Int, shrinkingFactor: Double) | ||
| extends UntypedQuantileNonSample(sketchSize, shrinkingFactor) with Serializable { | ||
| override def itemAsDouble(item: Any): Double = item.asInstanceOf[Byte].toDouble | ||
| } | ||
|
|
||
| @SerialVersionUID(1L) | ||
| class DoubleQuantileNonSample(sketchSize: Int, shrinkingFactor: Double) | ||
| extends UntypedQuantileNonSample(sketchSize, shrinkingFactor) with Serializable { | ||
| override def itemAsDouble(item: Any): Double = item.asInstanceOf[Double] | ||
| } | ||
|
|
||
| @SerialVersionUID(1L) | ||
| class FloatQuantileNonSample(sketchSize: Int, shrinkingFactor: Double) | ||
| extends UntypedQuantileNonSample(sketchSize, shrinkingFactor) with Serializable { | ||
| override def itemAsDouble(item: Any): Double = item.asInstanceOf[Float].toDouble | ||
| } | ||
|
|
||
| object KLLRunner { | ||
|
|
||
| def computeKLLSketchesInExtraPass( | ||
| data: DataFrame, | ||
| analyzers: Seq[Analyzer[State[_], Metric[_]]], | ||
| aggregateWith: Option[StateLoader] = None, | ||
| saveStatesTo: Option[StatePersister] = None) | ||
| : AnalyzerContext = { | ||
|
|
||
| val kllAnalyzers = analyzers.map { _.asInstanceOf[KLLSketch] } | ||
|
|
||
| val columnsAndParameters = kllAnalyzers | ||
| .map { analyzer => (analyzer.column, analyzer.kllParameters) } | ||
| .toMap | ||
|
|
||
| val sketching = sketchPartitions(columnsAndParameters, data.schema)_ | ||
|
|
||
| val sketchPerColumn = | ||
| data.rdd | ||
| .mapPartitions(sketching, preservesPartitioning = true) | ||
| .treeReduce { case (columnAndSketchesA, columnAndSketchesB) => | ||
| columnAndSketchesA.map { case (column, sketch) => | ||
| sketch.mergeUntyped(columnAndSketchesB(column)) | ||
| column -> sketch | ||
| } | ||
| } | ||
|
|
||
| val metricsByAnalyzer = kllAnalyzers.map { analyzer => | ||
| val kllState = sketchPerColumn(analyzer.column).asKLLState() | ||
| val metric = analyzer.calculateMetric(Some(kllState), aggregateWith, saveStatesTo) | ||
|
|
||
| analyzer -> metric | ||
| } | ||
|
|
||
| AnalyzerContext(metricsByAnalyzer.toMap[Analyzer[_, Metric[_]], Metric[_]]) | ||
| } | ||
|
|
||
| private[this] def emptySketches( | ||
| columnsAndParameters: Map[String, Option[KLLParameters]], | ||
| schema: StructType): Map[String, UntypedQuantileNonSample] = { | ||
|
|
||
| columnsAndParameters.map { case (column, parameters) => | ||
|
|
||
| val (sketchSize, shrinkingFactor) = parameters match { | ||
| case Some(kllParameters) => (kllParameters.sketchSize, kllParameters.shrinkingFactor) | ||
| case _ => (KLLSketch.DEFAULT_SKETCH_SIZE, KLLSketch.DEFAULT_SHRINKING_FACTOR) | ||
| } | ||
|
|
||
| val sketch: UntypedQuantileNonSample = schema(column).dataType match { | ||
| case DoubleType => new DoubleQuantileNonSample(sketchSize, shrinkingFactor) | ||
| case FloatType => new FloatQuantileNonSample(sketchSize, shrinkingFactor) | ||
| case ByteType => new ByteQuantileNonSample(sketchSize, shrinkingFactor) | ||
| case ShortType => new ShortQuantileNonSample(sketchSize, shrinkingFactor) | ||
| case IntegerType => new IntQuantileNonSample(sketchSize, shrinkingFactor) | ||
| case LongType => new LongQuantileNonSample(sketchSize, shrinkingFactor) | ||
| // TODO at the moment, we will throw exceptions for Decimals | ||
| case _ => throw new IllegalArgumentException(s"Cannot handle ${schema(column).dataType}") | ||
| } | ||
|
|
||
| column -> sketch | ||
| } | ||
| } | ||
|
|
||
| private[this] def sketchPartitions( | ||
| columnsAndParameters: Map[String, Option[KLLParameters]], | ||
| schema: StructType)(rows: Iterator[Row]) | ||
| : Iterator[Map[String, UntypedQuantileNonSample]] = { | ||
|
|
||
| val columnsAndSketches = emptySketches(columnsAndParameters, schema) | ||
|
|
||
| val namesToIndexes = schema.fields | ||
| .map { _.name } | ||
| .zipWithIndex | ||
| .toMap | ||
|
|
||
| // Include the index to avoid a lookup per row | ||
| val indexesAndSketches = columnsAndSketches.map { case (column, sketch) => | ||
| (namesToIndexes(column), sketch ) | ||
| } | ||
|
|
||
| while (rows.hasNext) { | ||
| val row = rows.next() | ||
| indexesAndSketches.foreach { case (index, sketch) => | ||
| if (!row.isNullAt(index)) { | ||
| sketch.updateUntyped(row.get(index)) | ||
| } | ||
| } | ||
| } | ||
|
|
||
| Iterator.single(columnsAndSketches) | ||
| } | ||
|
|
||
| } |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Oops, something went wrong.