add regression metrics

mllib/src/main/scala/org/apache/spark/mllib/evaluation/RegressionMetrics.scala

@@ -0,0 +1,83 @@
+/*
+ * 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.evaluation
+
+import org.apache.spark.annotation.Experimental
+import org.apache.spark.rdd.RDD
+import org.apache.spark.Logging
+import org.apache.spark.mllib.linalg.Vectors
+import org.apache.spark.mllib.stat.MultivariateOnlineSummarizer
+import org.apache.spark.mllib.rdd.RDDFunctions._
+
+/**
+ * :: Experimental ::
+ * Evaluator for regression.
+ *
+ * @param valuesAndPreds an RDD of (value, pred) pairs.
+ */
+@Experimental
+class RegressionMetrics(valuesAndPreds: RDD[(Double, Double)]) extends Logging {
+
+  /**
+   * Use MultivariateOnlineSummarizer to calculate mean and variance of different combination.
+   * MultivariateOnlineSummarizer is a numerically stable algorithm to compute mean and variance 
+   * in a online fashion.
+   */
+  private lazy val summarizer: MultivariateOnlineSummarizer = {
+    val summarizer: MultivariateOnlineSummarizer = valuesAndPreds.map{
+      case (value,pred) => Vectors.dense(
+        Array(value, pred, value - pred, math.abs(value - pred), math.pow(value - pred, 2.0))
+      )
+    }.treeAggregate(new MultivariateOnlineSummarizer())(
+        (summary, v) => summary.add(v),
+        (sum1,sum2) => sum1.merge(sum2)
+      )
+    summarizer
+  }
+
+  /**
+   * Computes the explained variance regression score
+   */
+  def explainedVarianceScore(): Double = {
+    1 - summarizer.variance(2) / summarizer.variance(0)
+  }
+
+  /**
+   * Computes the mean absolute error, which is a risk function corresponding to the
+   * expected value of the absolute error loss or l1-norm loss.
+   */
+  def mae(): Double = {
+    summarizer.mean(3)
+  }
+
+  /**
+   * Computes the mean square error, which is a risk function corresponding to the
+   * expected value of the squared error loss or quadratic loss.
+   */
+  def mse(): Double = {
+    summarizer.mean(4)
+  }
+
+  /**
+   * Computes R^2^, the coefficient of determination.
+   * @return
+   */
+  def r2_socre(): Double = {
+    1 - summarizer.mean(4) * summarizer.count / (summarizer.variance(0) * (summarizer.count - 1))
+  }
+}

mllib/src/test/scala/org/apache/spark/mllib/evaluation/RegressionMetricsSuite.scala

@@ -0,0 +1,45 @@
+/*
+ * 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.evaluation
+
+import org.scalatest.FunSuite
+import org.apache.spark.mllib.util.LocalSparkContext
+import org.apache.spark.mllib.util.TestingUtils._
+
+class RegressionMetricsSuite extends FunSuite with LocalSparkContext {
+
+  test("regression metrics") {
+    val valuesAndPreds = sc.parallelize(
+      Seq((3.0,2.5),(-0.5,0.0),(2.0,2.0),(7.0,8.0)),2)
+    val metrics = new RegressionMetrics(valuesAndPreds)
+    assert(metrics.explainedVarianceScore() ~== 0.95717 absTol 1E-5,"explained variance regression score mismatch")
+    assert(metrics.mae() ~== 0.5 absTol 1E-5, "mean absolute error mismatch")
+    assert(metrics.mse() ~== 0.375 absTol 1E-5, "mean square error mismatch")
+    assert(metrics.r2_socre() ~== 0.94861 absTol 1E-5, "r2 score mismatch")
+  }
+
+  test("regression metrics with complete fitting") {
+    val valuesAndPreds = sc.parallelize(
+      Seq((3.0,3.0),(0.0,0.0),(2.0,2.0),(8.0,8.0)),2)
+    val metrics = new RegressionMetrics(valuesAndPreds)
+    assert(metrics.explainedVarianceScore() ~== 1.0 absTol 1E-5,"explained variance regression score mismatch")
+    assert(metrics.mae() ~== 0.0 absTol 1E-5, "mean absolute error mismatch")
+    assert(metrics.mse() ~== 0.0 absTol 1E-5, "mean square error mismatch")
+    assert(metrics.r2_socre() ~== 1.0 absTol 1E-5, "r2 score mismatch")
+  }
+}