Tweaks to OpBinScoreEvaluator #233
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
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,237 @@ | ||
| /* | ||
| * Copyright (c) 2017, Salesforce.com, Inc. | ||
| * All rights reserved. | ||
| * | ||
| * Redistribution and use in source and binary forms, with or without | ||
| * modification, are permitted provided that the following conditions are met: | ||
| * | ||
| * * Redistributions of source code must retain the above copyright notice, this | ||
| * list of conditions and the following disclaimer. | ||
| * | ||
| * * Redistributions in binary form must reproduce the above copyright notice, | ||
| * this list of conditions and the following disclaimer in the documentation | ||
| * and/or other materials provided with the distribution. | ||
| * | ||
| * * Neither the name of the copyright holder nor the names of its | ||
| * contributors may be used to endorse or promote products derived from | ||
| * this software without specific prior written permission. | ||
| * | ||
| * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" | ||
| * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE | ||
| * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE | ||
| * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE | ||
| * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL | ||
| * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR | ||
| * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER | ||
| * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, | ||
| * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE | ||
| * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. | ||
| */ | ||
|
|
||
| package com.salesforce.op.evaluators | ||
|
|
||
| import org.apache.spark.rdd.RDD | ||
|
|
||
| /** | ||
| * Object to calculate Lift metrics for BinaryClassification problems | ||
| * Intended to build a Lift Plot. | ||
| * | ||
| * Algorithm for calculating a chart as seen here: | ||
| * https://www.kdnuggets.com/2016/03/lift-analysis-data-scientist-secret-weapon.html | ||
| */ | ||
| object LiftEvaluator { | ||
|
|
||
| /** | ||
| * Stores basic lift values for a specific band of scores | ||
| * | ||
| * @param group name / key for score band | ||
| * @param lowerBound minimum score represented in lift | ||
| * @param upperBound maximum score represented in lift | ||
| * @param rate optional calculated lift value, i.e. # yes / total count | ||
| * @param average optional lift rate across all score bands | ||
| * @param totalCount total number of records in score band | ||
| * @param yesCount number of yes records in score band | ||
| * @param noCount number of no records in score band | ||
| */ | ||
| case class LiftMetricBand | ||
| ( | ||
| group: String, | ||
| lowerBound: Double, | ||
| upperBound: Double, | ||
| rate: Option[Double], | ||
| average: Option[Double], | ||
| totalCount: Long, | ||
| yesCount: Long, | ||
| noCount: Long | ||
| ) extends EvaluationMetrics | ||
|
|
||
| /** | ||
| * Builds Seq[LiftMetricBand] for BinaryClassificationMetrics, calls liftMetricBands function | ||
| * with default score bands function | ||
| * | ||
| * @param scoreAndLabels RDD[(Double, Double)] of BinaryClassification (score, label) tuples | ||
| * @return Seq of LiftMetricBand containers of Lift calculations | ||
| */ | ||
| def apply | ||
| ( | ||
| scoreAndLabels: RDD[(Double, Double)] | ||
| ): Seq[LiftMetricBand] = { | ||
| liftMetricBands( | ||
| scoreAndLabels, | ||
| getDefaultScoreBands | ||
|
There was a problem hiding this comment. so why not allow users to customize score bands? |
||
| ) | ||
| } | ||
|
|
||
| /** | ||
| * Builds Seq of LiftMetricBand using RDD api | ||
| * | ||
| * @param scoreAndLabels RDD[(Double, Double)] of BinaryClassification (score, label) tuples | ||
| * @param getScoreBands function to calculate score bands, potentially using score distribution | ||
| * @return Seq of LiftMetricBand containers of Lift calculations | ||
| */ | ||
| private[op] def liftMetricBands | ||
| ( | ||
| scoreAndLabels: RDD[(Double, Double)], | ||
| getScoreBands: RDD[Double] => Seq[(Double, Double, String)] | ||
| ): Seq[LiftMetricBand] = { | ||
| val bands = getScoreBands(scoreAndLabels.map { case (score, _) => score }) | ||
|
There was a problem hiding this comment.
|
||
| val bandedLabels = scoreAndLabels.map { case (score, label) => | ||
| (categorizeScoreIntoBand((score, bands)), label) | ||
| }.collect { case (Some(band), label) => (band, label) } | ||
| val perBandCounts = aggregateBandedLabels(bandedLabels) | ||
| val overallRate = overallLiftRate(perBandCounts) | ||
| bands.map({ case (lower, upper, band) => | ||
|
There was a problem hiding this comment.
|
||
| formatLiftMetricBand(lower, upper, band, perBandCounts, overallRate) | ||
| }).sortBy(band => band.lowerBound) | ||
| } | ||
|
|
||
| /** | ||
| * function to return score bands for calculating lift | ||
| * Default: 10 equidistant bands for all 0.1 increments | ||
| * from 0.0 to 1.0 | ||
| * | ||
| * @param scores RDD of scores. unused in this function | ||
| * @return sequence of (lowerBound, upperBound, bandString) tuples | ||
| */ | ||
| private[op] def getDefaultScoreBands(scores: RDD[Double]): | ||
|
There was a problem hiding this comment. redundant endline |
||
| Seq[(Double, Double, String)] = | ||
| Seq( | ||
| (0.0, 0.1, "0-10"), | ||
| (0.1, 0.2, "10-20"), | ||
| (0.2, 0.3, "20-30"), | ||
| (0.3, 0.4, "30-40"), | ||
| (0.4, 0.5, "40-50"), | ||
| (0.5, 0.6, "50-60"), | ||
| (0.6, 0.7, "60-70"), | ||
| (0.7, 0.8, "70-80"), | ||
| (0.8, 0.9, "80-90"), | ||
| (0.9, 1.0, "90-100") | ||
| ) | ||
|
|
||
| /** | ||
| * PartialFunction. Defined when scores are [0.0, 1.0] | ||
| * Places a score Double into a score band based on | ||
| * lower and upper bounds | ||
| * | ||
| * @param score BinaryClassification score Double, [0.0, 1.0] | ||
| * @param bands sequence of upper/lower score bands | ||
| * @return optional key to describe categorized band, if found | ||
| */ | ||
| private[op] def categorizeScoreIntoBand: | ||
| PartialFunction[(Double, Seq[(Double, Double, String)]), Option[String]] = { | ||
| case (score: Double, bands: Seq[(Double, Double, String)]) | ||
| if (score >= 0.0) & (score <= 1.0) => | ||
| bands.find { case (l, u, _) => | ||
| (score >= l) & (score <= u) | ||
| } match { | ||
| case Some((_, _, bandString)) => Some(bandString) | ||
| case None => None | ||
| } | ||
| } | ||
|
|
||
| /** | ||
| * aggregates labels into counts by lift band | ||
| * | ||
| * @param bandedLabels PairRDD of (bandString, label) | ||
| * @return Map of bandString -> (total count, count of positive labels) | ||
| */ | ||
| private[op] def aggregateBandedLabels | ||
| ( | ||
| bandedLabels: RDD[(String, Double)] | ||
| ): Map[String, (Long, Long)] = { | ||
| val countsPerBand = bandedLabels.countByKey() | ||
| val truesPerBand = bandedLabels.aggregateByKey(zeroValue = 0.0)( | ||
| { case (sum, label) => sum + label }, | ||
| { case (sumX, sumY) => sumX + sumY }) | ||
| .collectAsMap() | ||
| countsPerBand.map { case (band, count) => | ||
| band -> (count, truesPerBand.getOrElse(band, 0.0).toLong) | ||
| }.toMap | ||
| } | ||
|
|
||
| /** | ||
| * calculates a baseline "yes" rate across score bands | ||
| * | ||
| * @param perBandCounts | ||
| * @return overall # yes / total records across all bands | ||
| */ | ||
| private[op] def overallLiftRate(perBandCounts: Map[String, (Long, Long)]): Option[Double] = { | ||
| val overallTotalCount = perBandCounts.values.map({ case (totalCount, _) => totalCount }).sum | ||
| val overallYesCount = perBandCounts.values.map({ case (_, yesCount) => yesCount }).sum | ||
| overallTotalCount match { | ||
| case 0L => None | ||
| case _ => Some(overallYesCount.toDouble / overallTotalCount) | ||
| } | ||
| } | ||
|
|
||
| /** | ||
| * Formats lift data in one band into LiftMetricBand data, | ||
| * including lower bound of score band, upper bound, total record | ||
| * count per band, and lift (# trues / total) | ||
| * | ||
| * @param lower lower bound of band | ||
| * @param upper upper bound of band | ||
| * @param bandString String key of band e.g. "10-20" | ||
| * @param perBandCounts calculated total counts and counts of true labels | ||
| * @param overallRate optional overall Lift rate across all bands | ||
| * @return LiftMetricBand container of metrics | ||
| */ | ||
| private[op] def formatLiftMetricBand | ||
| ( | ||
| lower: Double, | ||
| upper: Double, | ||
| bandString: String, | ||
| perBandCounts: Map[String, (Long, Long)], | ||
| overallRate: Option[Double] | ||
| ): LiftMetricBand = { | ||
| perBandCounts.get(bandString) match { | ||
| case Some((numTotal, numYes)) => { | ||
| val lift = numTotal match { | ||
| case 0L => None | ||
| case _ => Some(numYes.toDouble / numTotal) | ||
| } | ||
| LiftMetricBand( | ||
| group = bandString, | ||
| lowerBound = lower, | ||
| upperBound = upper, | ||
| rate = lift, | ||
| average = overallRate, | ||
| totalCount = numTotal, | ||
| yesCount = numYes, | ||
| noCount = numTotal - numYes | ||
| ) | ||
| } | ||
| case None => LiftMetricBand( | ||
| group = bandString, | ||
| lowerBound = lower, | ||
| upperBound = upper, | ||
| rate = None, | ||
| average = overallRate, | ||
| totalCount = 0L, | ||
| yesCount = 0L, | ||
| noCount = 0L | ||
| ) | ||
| } | ||
| } | ||
|
|
||
| } | ||
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.
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The evaluator has to extend
OpBinaryClassificationEvaluatorBase,OpMultiClassificationEvaluatorBaseorOpRegressionEvaluatorBase