[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 <meng@databricks.com>

Closes #13662 from mengxr/SPARK-15945.

2 files changed, 107 insertions, 2 deletions

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<Row> dataset = spark.createDataFrame(
+      Collections.singletonList(new LabeledPoint(1.0, x)), LabeledPoint.class
+    ).select("label", "features");
+    Dataset<Row> 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")
+    }
+  }
 }