Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[jvm-packages] unify the set features API (#7692)

Browse Source 
xgboost4j-spark provides 2 sets of API for setting features, one for CPU, another for GPU, which may cause confusion.

This PR removes the GPU API and adds an override CPU function setFeaturesCol to accept Array[String] parameters.
This commit is contained in:
Bobby Wang
2022-02-23 03:37:25 +08:00
committed by GitHub
parent c859764d29
commit e3e6de5ed9
6 changed files with 32 additions and 49 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
jvm-packages/xgboost4j-spark-gpu/src/test/scala/ml/dmlc/xgboost4j/scala/rapids/spark/GpuXGBoostClassifierSuite.scala
View File

@@ -1,5 +1,5 @@
/*
Copyright (c) 2021 by Contributors
Copyright (c) 2021-2022 by Contributors
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
@@ -39,7 +39,7 @@ class GpuXGBoostClassifierSuite extends GpuTestSuite {
StructField("f10", FloatType), StructField("f11", FloatType), StructField("f12", FloatType),
StructField(labelName, FloatType)
))
val featureNames = schema.fieldNames.filter(s => !s.equals(labelName)).toSeq
val featureNames = schema.fieldNames.filter(s => !s.equals(labelName))
test("The transform result should be same for several runs on same model") {
withGpuSparkSession(enableCsvConf()) { spark =>
@@ -90,7 +90,7 @@ class GpuXGBoostClassifierSuite extends GpuTestSuite {
.csv(dataPath).randomSplit(Array(0.7, 0.3), seed = 1)
val classifier = new XGBoostClassifier(xgbParam)
.setFeaturesCols(featureNames)
.setFeaturesCol(featureNames)
.setLabelCol(labelName)
.setTreeMethod("gpu_hist")
(classifier.fit(rawInput), testDf)
@@ -155,12 +155,12 @@ class GpuXGBoostClassifierSuite extends GpuTestSuite {
"please refer to setFeaturesCols"))
val left = cpuModel
.setFeaturesCols(featureNames)
.setFeaturesCol(featureNames)
.transform(testDf)
.collect()
val right = cpuModelFromFile
.setFeaturesCols(featureNames)
.setFeaturesCol(featureNames)
.transform(testDf)
.collect()
@@ -177,7 +177,7 @@ class GpuXGBoostClassifierSuite extends GpuTestSuite {
.csv(dataPath).randomSplit(Array(0.7, 0.3), seed = 1)
val classifier = new XGBoostClassifier(xgbParam)
.setFeaturesCols(featureNames)
.setFeaturesCol(featureNames)
.setLabelCol(labelName)
.setTreeMethod("gpu_hist")
classifier.fit(rawInput)

22
jvm-packages/xgboost4j-spark-gpu/src/test/scala/ml/dmlc/xgboost4j/scala/rapids/spark/GpuXGBoostGeneralSuite.scala
View File

@@ -1,5 +1,5 @@
/*
Copyright (c) 2021 by Contributors
Copyright (c) 2021-2022 by Contributors
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
@@ -28,7 +28,7 @@ class GpuXGBoostGeneralSuite extends GpuTestSuite {
private val labelName = "label_col"
private val weightName = "weight_col"
private val baseMarginName = "margin_col"
private val featureNames = Seq("f1", "f2", "f3")
private val featureNames = Array("f1", "f2", "f3")
private val allColumnNames = featureNames :+ weightName :+ baseMarginName :+ labelName
private val trainingData = Seq(
// f1, f2, f3, weight, margin, label
@@ -68,7 +68,7 @@ class GpuXGBoostGeneralSuite extends GpuTestSuite {
val xgbParam = Map("eta" -> 0.1f, "max_depth" -> 2, "objective" -> "multi:softprob",
"num_class" -> 3, "num_round" -> 5, "num_workers" -> 1, "tree_method" -> "gpu_hist")
new XGBoostClassifier(xgbParam)
.setFeaturesCols(featureNames)
.setFeaturesCol(featureNames)
.setLabelCol(labelName)
.fit(trainingDf)
}
@@ -84,7 +84,7 @@ class GpuXGBoostGeneralSuite extends GpuTestSuite {
"num_class" -> 3, "num_round" -> 5, "num_workers" -> 1, "tree_method" -> "gpu_hist")
val thrown1 = intercept[IllegalArgumentException] {
new XGBoostClassifier(xgbParam)
.setFeaturesCols(featureNames)
.setFeaturesCol(featureNames)
.setLabelCol(labelName)
.fit(trainingDf)
}
@@ -93,7 +93,7 @@ class GpuXGBoostGeneralSuite extends GpuTestSuite {
trainingDf = originalDf.withColumn(labelName, col(labelName).cast(StringType))
val thrown2 = intercept[IllegalArgumentException] {
new XGBoostClassifier(xgbParam)
.setFeaturesCols(featureNames)
.setFeaturesCol(featureNames)
.setLabelCol(labelName)
.fit(trainingDf)
}
@@ -117,7 +117,7 @@ class GpuXGBoostGeneralSuite extends GpuTestSuite {
val thrown1 = intercept[IllegalArgumentException] {
new XGBoostClassifier(xgbParam)
.setFeaturesCols(featureNames)
.setFeaturesCol(featureNames)
.fit(trainingDf)
}
assert(thrown1.getMessage.contains("label does not exist."))
@@ -132,7 +132,7 @@ class GpuXGBoostGeneralSuite extends GpuTestSuite {
"num_class" -> 3, "num_round" -> 5, "num_workers" -> 1, "tree_method" -> "hist")
val thrown = intercept[IllegalArgumentException] {
new XGBoostClassifier(xgbParam)
.setFeaturesCols(featureNames)
.setFeaturesCol(featureNames)
.setLabelCol(labelName)
.fit(trainingDf)
}
@@ -149,7 +149,7 @@ class GpuXGBoostGeneralSuite extends GpuTestSuite {
val xgbParam = Map("eta" -> 0.1f, "max_depth" -> 2, "objective" -> "multi:softprob",
"num_class" -> 3, "num_round" -> 5, "num_workers" -> 1, "tree_method" -> "gpu_hist")
val model1 = new XGBoostClassifier(xgbParam)
.setFeaturesCols(featureNames)
.setFeaturesCol(featureNames)
.setLabelCol(labelName)
.setEvalSets(Map("eval1" -> eval1, "eval2" -> eval2))
.fit(trainingDf)
@@ -166,7 +166,6 @@ class GpuXGBoostGeneralSuite extends GpuTestSuite {
test("test persistence of XGBoostClassifier and XGBoostClassificationModel") {
val xgbcPath = new File(tempDir.toFile, "xgbc").getPath
withGpuSparkSession() { spark =>
import spark.implicits._
val xgbParam = Map("eta" -> 0.1f, "max_depth" -> 2, "objective" -> "multi:softprob",
"num_class" -> 3, "num_round" -> 5, "num_workers" -> 1, "tree_method" -> "gpu_hist",
"features_cols" -> featureNames, "label_col" -> labelName)
@@ -174,7 +173,10 @@ class GpuXGBoostGeneralSuite extends GpuTestSuite {
xgbc.write.overwrite().save(xgbcPath)
val paramMap2 = XGBoostClassifier.load(xgbcPath).MLlib2XGBoostParams
xgbParam.foreach {
case (k, v) => assert(v.toString == paramMap2(k).toString)
case (k, v: Array[String]) =>
assert(v.sameElements(paramMap2(k).asInstanceOf[Array[String]]))
case (k, v) =>
assert(v.toString == paramMap2(k).toString)
}
}
}

2
jvm-packages/xgboost4j-spark-gpu/src/test/scala/ml/dmlc/xgboost4j/scala/rapids/spark/GpuXGBoostRegressorSuite.scala
View File

@@ -35,7 +35,7 @@ class GpuXGBoostRegressorSuite extends GpuTestSuite {
StructField("f3", FloatType),
StructField(groupName, IntegerType)))
val featureNames = schema.fieldNames.filter(s =>
!(s.equals(labelName) || s.equals(groupName))).toSeq
!(s.equals(labelName) || s.equals(groupName)))
test("The transform result should be same for several runs on same model") {
withGpuSparkSession(enableCsvConf()) { spark =>