weibozhao commented on code in PR #86: URL: https://github.com/apache/flink-ml/pull/86#discussion_r867650416
########## flink-ml-lib/src/main/java/org/apache/flink/ml/evaluation/binaryeval/BinaryClassificationEvaluator.java: ########## @@ -0,0 +1,783 @@ +/* + * 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.flink.ml.evaluation.binaryeval; + +import org.apache.flink.api.common.functions.MapFunction; +import org.apache.flink.api.common.functions.MapPartitionFunction; +import org.apache.flink.api.common.functions.RichFlatMapFunction; +import org.apache.flink.api.common.functions.RichMapFunction; +import org.apache.flink.api.common.functions.RichMapPartitionFunction; +import org.apache.flink.api.common.state.ListState; +import org.apache.flink.api.common.state.ListStateDescriptor; +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.api.java.tuple.Tuple3; +import org.apache.flink.api.java.tuple.Tuple4; +import org.apache.flink.api.java.typeutils.RowTypeInfo; +import org.apache.flink.api.scala.typeutils.Types; +import org.apache.flink.iteration.operator.OperatorStateUtils; +import org.apache.flink.ml.api.AlgoOperator; +import org.apache.flink.ml.common.broadcast.BroadcastUtils; +import org.apache.flink.ml.common.datastream.DataStreamUtils; +import org.apache.flink.ml.linalg.DenseVector; +import org.apache.flink.ml.param.Param; +import org.apache.flink.ml.util.ParamUtils; +import org.apache.flink.ml.util.ReadWriteUtils; +import org.apache.flink.runtime.state.StateInitializationContext; +import org.apache.flink.runtime.state.StateSnapshotContext; +import org.apache.flink.streaming.api.datastream.DataStream; +import org.apache.flink.streaming.api.operators.AbstractStreamOperator; +import org.apache.flink.streaming.api.operators.BoundedOneInput; +import org.apache.flink.streaming.api.operators.OneInputStreamOperator; +import org.apache.flink.streaming.api.operators.StreamMap; +import org.apache.flink.streaming.api.watermark.Watermark; +import org.apache.flink.streaming.runtime.streamrecord.StreamRecord; +import org.apache.flink.table.api.Table; +import org.apache.flink.table.api.bridge.java.StreamTableEnvironment; +import org.apache.flink.table.api.internal.TableImpl; +import org.apache.flink.types.Row; +import org.apache.flink.util.Collector; +import org.apache.flink.util.Preconditions; + +import java.io.IOException; +import java.io.Serializable; +import java.util.ArrayList; +import java.util.Arrays; +import java.util.Collections; +import java.util.Comparator; +import java.util.HashMap; +import java.util.Iterator; +import java.util.LinkedList; +import java.util.List; +import java.util.Map; +import java.util.Random; + +import static org.apache.flink.runtime.blob.BlobWriter.LOG; + +/** + * Calculates the evaluation metrics for binary classification. The input data has columns + * rawPrediction, label and an optional weight column. The rawPrediction can be of type double + * (binary 0/1 prediction, or probability of label 1) or of type vector (length-2 vector of raw + * predictions, scores, or label probabilities). The output may contain different metrics which will + * be defined by parameter MetricsNames. See @BinaryClassificationEvaluatorParams. + */ +public class BinaryClassificationEvaluator + implements AlgoOperator<BinaryClassificationEvaluator>, + BinaryClassificationEvaluatorParams<BinaryClassificationEvaluator> { + private final Map<Param<?>, Object> paramMap = new HashMap<>(); + private static final int NUM_SAMPLE_FOR_RANGE_PARTITION = 100; + + public BinaryClassificationEvaluator() { + ParamUtils.initializeMapWithDefaultValues(paramMap, this); + } + + @Override + @SuppressWarnings("unchecked") + public Table[] transform(Table... inputs) { + Preconditions.checkArgument(inputs.length == 1); + StreamTableEnvironment tEnv = + (StreamTableEnvironment) ((TableImpl) inputs[0]).getTableEnvironment(); + DataStream<Tuple3<Double, Boolean, Double>> evalData = + tEnv.toDataStream(inputs[0]) + .map(new ParseSample(getLabelCol(), getRawPredictionCol(), getWeightCol())); + final String boundaryRangeKey = "boundaryRange"; + final String partitionSummariesKey = "partitionSummaries"; + + DataStream<Tuple4<Double, Boolean, Double, Integer>> evalDataWithTaskId = + BroadcastUtils.withBroadcastStream( + Collections.singletonList(evalData), + Collections.singletonMap(boundaryRangeKey, getBoundaryRange(evalData)), + inputList -> { + DataStream input = inputList.get(0); + return input.map(new AppendTaskId(boundaryRangeKey)); + }); + + /* Repartition the evaluated data by range. */ + evalDataWithTaskId = + evalDataWithTaskId.partitionCustom((chunkId, numPartitions) -> chunkId, x -> x.f3); + + /* Sorts local data by score.*/ + evalData = + DataStreamUtils.mapPartition( + evalDataWithTaskId, + new MapPartitionFunction< + Tuple4<Double, Boolean, Double, Integer>, + Tuple3<Double, Boolean, Double>>() { + @Override + public void mapPartition( + Iterable<Tuple4<Double, Boolean, Double, Integer>> values, + Collector<Tuple3<Double, Boolean, Double>> out) { + List<Tuple3<Double, Boolean, Double>> bufferedData = + new LinkedList<>(); + for (Tuple4<Double, Boolean, Double, Integer> t4 : values) { + bufferedData.add(Tuple3.of(t4.f0, t4.f1, t4.f2)); + } + bufferedData.sort(Comparator.comparingDouble(o -> -o.f0)); + for (Tuple3<Double, Boolean, Double> dataPoint : bufferedData) { + out.collect(dataPoint); + } + } + }); + + /* Calculates the summary of local data. */ + DataStream<BinarySummary> partitionSummaries = + evalData.transform( + "reduceInEachPartition", + TypeInformation.of(BinarySummary.class), + new PartitionSummaryOperator()); + + /* Sorts global data. Output Tuple4 : <score, order, isPositive, weight> */ + DataStream<Tuple4<Double, Long, Boolean, Double>> dataWithOrders = + BroadcastUtils.withBroadcastStream( + Collections.singletonList(evalData), + Collections.singletonMap(partitionSummariesKey, partitionSummaries), + inputList -> { + DataStream input = inputList.get(0); + return input.flatMap(new CalcSampleOrders(partitionSummariesKey)); + }); + + dataWithOrders = + dataWithOrders.transform( + "appendMaxWaterMark", + dataWithOrders.getType(), + new AppendMaxWatermark(x -> x)); + + DataStream<double[]> localAucVariable = + dataWithOrders.transform( + "AccumulateMultiScore", + TypeInformation.of(double[].class), + new AccumulateMultiScoreOperator()); + + DataStream<double[]> middleAreaUnderROC = + localAucVariable + .transform( + "calcLocalAucValues", + TypeInformation.of(double[].class), + new AucOperator()) + .transform( + "calcGlobalAucValues", + TypeInformation.of(double[].class), + new AucOperator()) + .setParallelism(1); + + DataStream<Double> areaUnderROC = + middleAreaUnderROC.map( + (MapFunction<double[], Double>) + value -> { + if (value[1] > 0 && value[2] > 0) { + return (value[0] - 1. * value[1] * (value[1] + 1) / 2) Review Comment: some value in doubleArray has no meaning, just a middle value in calculating AUC. If you want to know the meaning of variables, you can read the code which calculate these variables. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@flink.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org