From 63e0aebe22ba41c636ecaddd8647721d7690a1ec Mon Sep 17 00:00:00 2001 From: Xiangrui Meng Date: Tue, 14 Jun 2016 18:57:45 -0700 Subject: [SPARK-15945][MLLIB] Conversion between old/new vector columns in a DataFrame (Scala/Java) ## What changes were proposed in this pull request? This PR provides conversion utils between old/new vector columns in a DataFrame. So users can use it to migrate their datasets and pipelines manually. The methods are implemented under `MLUtils` and called `convertVectorColumnsToML` and `convertVectorColumnsFromML`. Both take a DataFrame and a list of vector columns to be converted. It is a no-op on vector columns that are already converted. A warning message is logged if actual conversion happens. This is the first sub-task under SPARK-15944 to make it easier to migrate existing pipelines to Spark 2.0. ## How was this patch tested? Unit tests in Scala and Java. cc: yanboliang Author: Xiangrui Meng Closes #13662 from mengxr/SPARK-15945. --- .../org/apache/spark/mllib/util/MLUtils.scala | 117 +++++++++++++++++++-- .../apache/spark/mllib/util/JavaMLUtilsSuite.java | 49 +++++++++ .../org/apache/spark/mllib/util/MLUtilsSuite.scala | 60 ++++++++++- 3 files changed, 218 insertions(+), 8 deletions(-) create mode 100644 mllib/src/test/java/org/apache/spark/mllib/util/JavaMLUtilsSuite.java (limited to 'mllib/src') diff --git a/mllib/src/main/scala/org/apache/spark/mllib/util/MLUtils.scala b/mllib/src/main/scala/org/apache/spark/mllib/util/MLUtils.scala index f0346e600d..7d5bdffc42 100644 --- a/mllib/src/main/scala/org/apache/spark/mllib/util/MLUtils.scala +++ b/mllib/src/main/scala/org/apache/spark/mllib/util/MLUtils.scala @@ -17,14 +17,19 @@ package org.apache.spark.mllib.util +import scala.annotation.varargs import scala.reflect.ClassTag import org.apache.spark.SparkContext import org.apache.spark.annotation.Since -import org.apache.spark.mllib.linalg.{DenseVector, SparseVector, Vector, Vectors} +import org.apache.spark.internal.Logging +import org.apache.spark.ml.linalg.{VectorUDT => MLVectorUDT} +import org.apache.spark.mllib.linalg._ import org.apache.spark.mllib.linalg.BLAS.dot import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.rdd.{PartitionwiseSampledRDD, RDD} +import org.apache.spark.sql.{DataFrame, Dataset} +import org.apache.spark.sql.functions.{col, udf} import org.apache.spark.storage.StorageLevel import org.apache.spark.util.random.BernoulliCellSampler @@ -32,7 +37,7 @@ import org.apache.spark.util.random.BernoulliCellSampler * Helper methods to load, save and pre-process data used in ML Lib. */ @Since("0.8.0") -object MLUtils { +object MLUtils extends Logging { private[mllib] lazy val EPSILON = { var eps = 1.0 @@ -50,7 +55,6 @@ object MLUtils { * where the indices are one-based and in ascending order. * This method parses each line into a [[org.apache.spark.mllib.regression.LabeledPoint]], * where the feature indices are converted to zero-based. - * * @param sc Spark context * @param path file or directory path in any Hadoop-supported file system URI * @param numFeatures number of features, which will be determined from the input data if a @@ -145,7 +149,6 @@ object MLUtils { * Save labeled data in LIBSVM format. * @param data an RDD of LabeledPoint to be saved * @param dir directory to save the data - * * @see [[org.apache.spark.mllib.util.MLUtils#loadLibSVMFile]] */ @Since("1.0.0") @@ -253,6 +256,110 @@ object MLUtils { } } + /** + * Converts vector columns in an input Dataset from the [[org.apache.spark.mllib.linalg.Vector]] + * type to the new [[org.apache.spark.ml.linalg.Vector]] type under the `spark.ml` package. + * @param dataset input dataset + * @param cols a list of vector columns to be converted. New vector columns will be ignored. If + * unspecified, all old vector columns will be converted except nested ones. + * @return the input [[DataFrame]] with old vector columns converted to the new vector type + */ + @Since("2.0.0") + @varargs + def convertVectorColumnsToML(dataset: Dataset[_], cols: String*): DataFrame = { + val schema = dataset.schema + val colSet = if (cols.nonEmpty) { + cols.flatMap { c => + val dataType = schema(c).dataType + if (dataType.getClass == classOf[VectorUDT]) { + Some(c) + } else { + // ignore new vector columns and raise an exception on other column types + require(dataType.getClass == classOf[MLVectorUDT], + s"Column $c must be old Vector type to be converted to new type but got $dataType.") + None + } + }.toSet + } else { + schema.fields + .filter(_.dataType.getClass == classOf[VectorUDT]) + .map(_.name) + .toSet + } + + if (colSet.isEmpty) { + return dataset.toDF() + } + + logWarning("Vector column conversion has serialization overhead. " + + "Please migrate your datasets and workflows to use the spark.ml package.") + + // TODO: This implementation has performance issues due to unnecessary serialization. + // TODO: It is better (but trickier) if we can cast the old vector type to new type directly. + val convertToML = udf { v: Vector => v.asML } + val exprs = schema.fields.map { field => + val c = field.name + if (colSet.contains(c)) { + convertToML(col(c)).as(c, field.metadata) + } else { + col(c) + } + } + dataset.select(exprs: _*) + } + + /** + * Converts vector columns in an input Dataset to the [[org.apache.spark.ml.linalg.Vector]] type + * from the new [[org.apache.spark.mllib.linalg.Vector]] type under the `spark.ml` package. + * @param dataset input dataset + * @param cols a list of vector columns to be converted. Old vector columns will be ignored. If + * unspecified, all new vector columns will be converted except nested ones. + * @return the input [[DataFrame]] with new vector columns converted to the old vector type + */ + @Since("2.0.0") + @varargs + def convertVectorColumnsFromML(dataset: Dataset[_], cols: String*): DataFrame = { + val schema = dataset.schema + val colSet = if (cols.nonEmpty) { + cols.flatMap { c => + val dataType = schema(c).dataType + if (dataType.getClass == classOf[MLVectorUDT]) { + Some(c) + } else { + // ignore old vector columns and raise an exception on other column types + require(dataType.getClass == classOf[VectorUDT], + s"Column $c must be new Vector type to be converted to old type but got $dataType.") + None + } + }.toSet + } else { + schema.fields + .filter(_.dataType.getClass == classOf[MLVectorUDT]) + .map(_.name) + .toSet + } + + if (colSet.isEmpty) { + return dataset.toDF() + } + + logWarning("Vector column conversion has serialization overhead. " + + "Please migrate your datasets and workflows to use the spark.ml package.") + + // TODO: This implementation has performance issues due to unnecessary serialization. + // TODO: It is better (but trickier) if we can cast the new vector type to old type directly. + val convertFromML = udf { Vectors.fromML _ } + val exprs = schema.fields.map { field => + val c = field.name + if (colSet.contains(c)) { + convertFromML(col(c)).as(c, field.metadata) + } else { + col(c) + } + } + dataset.select(exprs: _*) + } + /** * Returns the squared Euclidean distance between two vectors. The following formula will be used * if it does not introduce too much numerical error: @@ -261,7 +368,6 @@ object MLUtils { * * When both vector norms are given, this is faster than computing the squared distance directly, * especially when one of the vectors is a sparse vector. - * * @param v1 the first vector * @param norm1 the norm of the first vector, non-negative * @param v2 the second vector @@ -314,7 +420,6 @@ object MLUtils { * When `x` is positive and large, computing `math.log(1 + math.exp(x))` will lead to arithmetic * overflow. This will happen when `x > 709.78` which is not a very large number. * It can be addressed by rewriting the formula into `x + math.log1p(math.exp(-x))` when `x > 0`. - * * @param x a floating-point value as input. * @return the result of `math.log(1 + math.exp(x))`. */ diff --git a/mllib/src/test/java/org/apache/spark/mllib/util/JavaMLUtilsSuite.java b/mllib/src/test/java/org/apache/spark/mllib/util/JavaMLUtilsSuite.java new file mode 100644 index 0000000000..2fa0bd2546 --- /dev/null +++ b/mllib/src/test/java/org/apache/spark/mllib/util/JavaMLUtilsSuite.java @@ -0,0 +1,49 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License 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 org.apache.spark.mllib.util; + +import java.util.Collections; + +import org.junit.Assert; +import org.junit.Test; + +import org.apache.spark.SharedSparkSession; +import org.apache.spark.mllib.linalg.Vector; +import org.apache.spark.mllib.linalg.Vectors; +import org.apache.spark.mllib.regression.LabeledPoint; +import org.apache.spark.sql.Dataset; +import org.apache.spark.sql.Row; +import org.apache.spark.sql.RowFactory; + +public class JavaMLUtilsSuite extends SharedSparkSession { + + @Test + public void testConvertVectorColumnsToAndFromML() { + Vector x = Vectors.dense(2.0); + Dataset dataset = spark.createDataFrame( + Collections.singletonList(new LabeledPoint(1.0, x)), LabeledPoint.class + ).select("label", "features"); + Dataset newDataset1 = MLUtils.convertVectorColumnsToML(dataset); + Row new1 = newDataset1.first(); + Assert.assertEquals(RowFactory.create(1.0, x.asML()), new1); + Row new2 = MLUtils.convertVectorColumnsToML(dataset, "features").first(); + Assert.assertEquals(new1, new2); + Row old1 = MLUtils.convertVectorColumnsFromML(newDataset1).first(); + Assert.assertEquals(RowFactory.create(1.0, x), old1); + } +} diff --git a/mllib/src/test/scala/org/apache/spark/mllib/util/MLUtilsSuite.scala b/mllib/src/test/scala/org/apache/spark/mllib/util/MLUtilsSuite.scala index 7b6bfee00c..3801bd127a 100644 --- a/mllib/src/test/scala/org/apache/spark/mllib/util/MLUtilsSuite.scala +++ b/mllib/src/test/scala/org/apache/spark/mllib/util/MLUtilsSuite.scala @@ -25,12 +25,14 @@ import scala.io.Source import breeze.linalg.{squaredDistance => breezeSquaredDistance} import com.google.common.io.Files -import org.apache.spark.SparkException -import org.apache.spark.SparkFunSuite +import org.apache.spark.{SparkException, SparkFunSuite} import org.apache.spark.mllib.linalg.{DenseVector, SparseVector, Vectors} import org.apache.spark.mllib.regression.LabeledPoint import org.apache.spark.mllib.util.MLUtils._ import org.apache.spark.mllib.util.TestingUtils._ +import org.apache.spark.sql.Row +import org.apache.spark.sql.functions.col +import org.apache.spark.sql.types.MetadataBuilder import org.apache.spark.util.Utils class MLUtilsSuite extends SparkFunSuite with MLlibTestSparkContext { @@ -245,4 +247,58 @@ class MLUtilsSuite extends SparkFunSuite with MLlibTestSparkContext { assert(log1pExp(-13.8) ~== math.log1p(math.exp(-13.8)) absTol 1E-10) assert(log1pExp(-238423789.865) ~== math.log1p(math.exp(-238423789.865)) absTol 1E-10) } + + test("convertVectorColumnsToML") { + val x = Vectors.sparse(2, Array(1), Array(1.0)) + val metadata = new MetadataBuilder().putLong("numFeatures", 2L).build() + val y = Vectors.dense(2.0, 3.0) + val z = Vectors.dense(4.0) + val p = (5.0, z) + val w = Vectors.dense(6.0).asML + val df = spark.createDataFrame(Seq( + (0, x, y, p, w) + )).toDF("id", "x", "y", "p", "w") + .withColumn("x", col("x"), metadata) + val newDF1 = convertVectorColumnsToML(df) + assert(newDF1.schema("x").metadata === metadata, "Metadata should be preserved.") + val new1 = newDF1.first() + assert(new1 === Row(0, x.asML, y.asML, Row(5.0, z), w)) + val new2 = convertVectorColumnsToML(df, "x", "y").first() + assert(new2 === new1) + val new3 = convertVectorColumnsToML(df, "y", "w").first() + assert(new3 === Row(0, x, y.asML, Row(5.0, z), w)) + intercept[IllegalArgumentException] { + convertVectorColumnsToML(df, "p") + } + intercept[IllegalArgumentException] { + convertVectorColumnsToML(df, "p._2") + } + } + + test("convertVectorColumnsFromML") { + val x = Vectors.sparse(2, Array(1), Array(1.0)).asML + val metadata = new MetadataBuilder().putLong("numFeatures", 2L).build() + val y = Vectors.dense(2.0, 3.0).asML + val z = Vectors.dense(4.0).asML + val p = (5.0, z) + val w = Vectors.dense(6.0) + val df = spark.createDataFrame(Seq( + (0, x, y, p, w) + )).toDF("id", "x", "y", "p", "w") + .withColumn("x", col("x"), metadata) + val newDF1 = convertVectorColumnsFromML(df) + assert(newDF1.schema("x").metadata === metadata, "Metadata should be preserved.") + val new1 = newDF1.first() + assert(new1 === Row(0, Vectors.fromML(x), Vectors.fromML(y), Row(5.0, z), w)) + val new2 = convertVectorColumnsFromML(df, "x", "y").first() + assert(new2 === new1) + val new3 = convertVectorColumnsFromML(df, "y", "w").first() + assert(new3 === Row(0, x, Vectors.fromML(y), Row(5.0, z), w)) + intercept[IllegalArgumentException] { + convertVectorColumnsFromML(df, "p") + } + intercept[IllegalArgumentException] { + convertVectorColumnsFromML(df, "p._2") + } + } } -- cgit v1.2.3