Github user mengxr commented on a diff in the pull request: https://github.com/apache/spark/pull/3519#discussion_r23275412 --- Diff: mllib/src/main/scala/org/apache/spark/mllib/regression/IsotonicRegression.scala --- @@ -0,0 +1,235 @@ +/* + * 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.regression + +import org.apache.spark.api.java.{JavaRDD, JavaPairRDD} +import org.apache.spark.rdd.RDD + +/** + * Regression model for Isotonic regression + * + * @param predictions Weights computed for every feature. + * @param isotonic isotonic (increasing) or antitonic (decreasing) sequence + */ +class IsotonicRegressionModel ( + val predictions: Seq[(Double, Double, Double)], + val isotonic: Boolean) + extends Serializable { + + /** + * Predict labels for provided features + * + * @param testData features to be labeled + * @return predicted labels + */ + def predict(testData: RDD[Double]): RDD[Double] = + testData.map(predict) + + /** + * Predict labels for provided features + * + * @param testData features to be labeled + * @return predicted labels + */ + def predict(testData: JavaRDD[java.lang.Double]): JavaRDD[java.lang.Double] = + testData.rdd.map(_.doubleValue()).map(predict).map(new java.lang.Double(_)) + + /** + * Predict a single label + * + * @param testData feature to be labeled + * @return predicted label + */ + def predict(testData: Double): Double = + // Take the highest of data points smaller than our feature or data point with lowest feature + (predictions.head +: predictions.filter(y => y._2 <= testData)).last._1 +} + +/** + * Base representing algorithm for isotonic regression + */ +trait IsotonicRegressionAlgorithm + extends Serializable { + + /** + * Creates isotonic regression model with given parameters + * + * @param predictions labels estimated using isotonic regression algorithm. + * Used for predictions on new data points. + * @param isotonic isotonic (increasing) or antitonic (decreasing) sequence + * @return isotonic regression model + */ + protected def createModel( + predictions: Seq[(Double, Double, Double)], + isotonic: Boolean): IsotonicRegressionModel + + /** + * Run algorithm to obtain isotonic regression model + * + * @param input (label, feature, weight) + * @param isotonic isotonic (increasing) or antitonic (decreasing) sequence + * @return isotonic regression model + */ + def run( + input: RDD[(Double, Double, Double)], + isotonic: Boolean): IsotonicRegressionModel +} + +/** + * Parallel pool adjacent violators algorithm for monotone regression + */ +class PoolAdjacentViolators private [mllib] + extends IsotonicRegressionAlgorithm { + + override def run( + input: RDD[(Double, Double, Double)], + isotonic: Boolean = true): IsotonicRegressionModel = { + createModel( + parallelPoolAdjacentViolators(input, isotonic), + isotonic) + } + + override protected def createModel( + predictions: Seq[(Double, Double, Double)], + isotonic: Boolean): IsotonicRegressionModel = { + new IsotonicRegressionModel(predictions, isotonic) + } + + /** + * Performs a pool adjacent violators algorithm (PAVA) + * Uses approach with single processing of data where violators in previously processed + * data created by pooling are fixed immediatelly. + * Uses optimization of discovering monotonicity violating sequences + * Method in situ mutates input array + * + * @param in input data + * @param isotonic asc or desc + * @return result + */ + private def poolAdjacentViolators( + in: Array[(Double, Double, Double)], + isotonic: Boolean): Array[(Double, Double, Double)] = { + + // Pools sub array within given bounds assigning weighted average value to all elements + def pool(in: Array[(Double, Double, Double)], start: Int, end: Int): Unit = { + val poolSubArray = in.slice(start, end + 1) + + val weightedSum = poolSubArray.map(lp => lp._1 * lp._3).sum + val weight = poolSubArray.map(_._3).sum + + for(i <- start to end) { + in(i) = (weightedSum / weight, in(i)._2, in(i)._3) + } + } + + val isotonicConstraint: (Double, Double) => Boolean = (x, y) => x <= y + val antitonicConstraint: (Double, Double) => Boolean = (x, y) => x >= y + + def monotonicityConstraint(isotonic: Boolean) = + if(isotonic) isotonicConstraint else antitonicConstraint + + val monotonicityConstraintHolds = monotonicityConstraint(isotonic) + + var i = 0 + + while(i < in.length) { --- End diff -- space before `(`
--- If your project is set up for it, you can reply to this email and have your reply appear on GitHub as well. If your project does not have this feature enabled and wishes so, or if the feature is enabled but not working, please contact infrastructure at infrastruct...@apache.org or file a JIRA ticket with INFRA. --- --------------------------------------------------------------------- To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org