examples/ml/JavaModelSelectionViaCrossValidationExample.java

0001 /*
0002  * Licensed to the Apache Software Foundation (ASF) under one or more
0003  * contributor license agreements.  See the NOTICE file distributed with
0004  * this work for additional information regarding copyright ownership.
0005  * The ASF licenses this file to You under the Apache License, Version 2.0
0006  * (the "License"); you may not use this file except in compliance with
0007  * the License.  You may obtain a copy of the License at
0008  *
0009  *    http://www.apache.org/licenses/LICENSE-2.0
0010  *
0011  * Unless required by applicable law or agreed to in writing, software
0012  * distributed under the License is distributed on an "AS IS" BASIS,
0013  * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
0014  * See the License for the specific language governing permissions and
0015  * limitations under the License.
0016  */
0017
0018 package org.apache.spark.examples.ml;
0019
0020 // $example on$
0021 import java.util.Arrays;
0022 // $example off$
0023
0024 // $example on$
0025 import org.apache.spark.ml.Pipeline;
0026 import org.apache.spark.ml.PipelineStage;
0027 import org.apache.spark.ml.classification.LogisticRegression;
0028 import org.apache.spark.ml.evaluation.BinaryClassificationEvaluator;
0029 import org.apache.spark.ml.feature.HashingTF;
0030 import org.apache.spark.ml.feature.Tokenizer;
0031 import org.apache.spark.ml.param.ParamMap;
0032 import org.apache.spark.ml.tuning.CrossValidator;
0033 import org.apache.spark.ml.tuning.CrossValidatorModel;
0034 import org.apache.spark.ml.tuning.ParamGridBuilder;
0035 import org.apache.spark.sql.Dataset;
0036 import org.apache.spark.sql.Row;
0037 // $example off$
0038 import org.apache.spark.sql.SparkSession;
0039
0040 /**
0041  * Java example for Model Selection via Cross Validation.
0042  */
0043 public class JavaModelSelectionViaCrossValidationExample {
0044   public static void main(String[] args) {
0045     SparkSession spark = SparkSession
0046       .builder()
0047       .appName("JavaModelSelectionViaCrossValidationExample")
0048       .getOrCreate();
0049
0050     // $example on$
0051     // Prepare training documents, which are labeled.
0052     Dataset<Row> training = spark.createDataFrame(Arrays.asList(
0053       new JavaLabeledDocument(0L, "a b c d e spark", 1.0),
0054       new JavaLabeledDocument(1L, "b d", 0.0),
0055       new JavaLabeledDocument(2L,"spark f g h", 1.0),
0056       new JavaLabeledDocument(3L, "hadoop mapreduce", 0.0),
0057       new JavaLabeledDocument(4L, "b spark who", 1.0),
0058       new JavaLabeledDocument(5L, "g d a y", 0.0),
0059       new JavaLabeledDocument(6L, "spark fly", 1.0),
0060       new JavaLabeledDocument(7L, "was mapreduce", 0.0),
0061       new JavaLabeledDocument(8L, "e spark program", 1.0),
0062       new JavaLabeledDocument(9L, "a e c l", 0.0),
0063       new JavaLabeledDocument(10L, "spark compile", 1.0),
0064       new JavaLabeledDocument(11L, "hadoop software", 0.0)
0065     ), JavaLabeledDocument.class);
0066
0067     // Configure an ML pipeline, which consists of three stages: tokenizer, hashingTF, and lr.
0068     Tokenizer tokenizer = new Tokenizer()
0069       .setInputCol("text")
0070       .setOutputCol("words");
0071     HashingTF hashingTF = new HashingTF()
0072       .setNumFeatures(1000)
0073       .setInputCol(tokenizer.getOutputCol())
0074       .setOutputCol("features");
0075     LogisticRegression lr = new LogisticRegression()
0076       .setMaxIter(10)
0077       .setRegParam(0.01);
0078     Pipeline pipeline = new Pipeline()
0079       .setStages(new PipelineStage[] {tokenizer, hashingTF, lr});
0080
0081     // We use a ParamGridBuilder to construct a grid of parameters to search over.
0082     // With 3 values for hashingTF.numFeatures and 2 values for lr.regParam,
0083     // this grid will have 3 x 2 = 6 parameter settings for CrossValidator to choose from.
0084     ParamMap[] paramGrid = new ParamGridBuilder()
0085       .addGrid(hashingTF.numFeatures(), new int[] {10, 100, 1000})
0086       .addGrid(lr.regParam(), new double[] {0.1, 0.01})
0087       .build();
0088
0089     // We now treat the Pipeline as an Estimator, wrapping it in a CrossValidator instance.
0090     // This will allow us to jointly choose parameters for all Pipeline stages.
0091     // A CrossValidator requires an Estimator, a set of Estimator ParamMaps, and an Evaluator.
0092     // Note that the evaluator here is a BinaryClassificationEvaluator and its default metric
0093     // is areaUnderROC.
0094     CrossValidator cv = new CrossValidator()
0095       .setEstimator(pipeline)
0096       .setEvaluator(new BinaryClassificationEvaluator())
0097       .setEstimatorParamMaps(paramGrid)
0098       .setNumFolds(2)  // Use 3+ in practice
0099       .setParallelism(2);  // Evaluate up to 2 parameter settings in parallel
0100
0101     // Run cross-validation, and choose the best set of parameters.
0102     CrossValidatorModel cvModel = cv.fit(training);
0103
0104     // Prepare test documents, which are unlabeled.
0105     Dataset<Row> test = spark.createDataFrame(Arrays.asList(
0106       new JavaDocument(4L, "spark i j k"),
0107       new JavaDocument(5L, "l m n"),
0108       new JavaDocument(6L, "mapreduce spark"),
0109       new JavaDocument(7L, "apache hadoop")
0110     ), JavaDocument.class);
0111
0112     // Make predictions on test documents. cvModel uses the best model found (lrModel).
0113     Dataset<Row> predictions = cvModel.transform(test);
0114     for (Row r : predictions.select("id", "text", "probability", "prediction").collectAsList()) {
0115       System.out.println("(" + r.get(0) + ", " + r.get(1) + ") --> prob=" + r.get(2)
0116         + ", prediction=" + r.get(3));
0117     }
0118     // $example off$
0119
0120     spark.stop();
0121   }
0122 }