zhipeng93 commented on a change in pull request #28: URL: https://github.com/apache/flink-ml/pull/28#discussion_r762762819
########## File path: flink-ml-lib/src/main/java/org/apache/flink/ml/classification/linear/LogisticRegressionModel.java ########## @@ -0,0 +1,185 @@ +/* + * 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.classification.linear; + +import org.apache.flink.api.common.functions.AbstractRichFunction; +import org.apache.flink.api.common.typeinfo.BasicTypeInfo; +import org.apache.flink.api.common.typeinfo.TypeInformation; +import org.apache.flink.api.java.tuple.Tuple2; +import org.apache.flink.api.java.typeutils.RowTypeInfo; +import org.apache.flink.ml.api.Model; +import org.apache.flink.ml.common.broadcast.BroadcastUtils; +import org.apache.flink.ml.common.datastream.TableUtils; +import org.apache.flink.ml.linalg.BLAS; +import org.apache.flink.ml.linalg.DenseVector; +import org.apache.flink.ml.linalg.Vectors; +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.streaming.api.datastream.DataStream; +import org.apache.flink.streaming.api.environment.StreamExecutionEnvironment; +import org.apache.flink.streaming.api.operators.AbstractUdfStreamOperator; +import org.apache.flink.streaming.api.operators.OneInputStreamOperator; +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.Preconditions; + +import org.apache.commons.lang3.ArrayUtils; + +import java.io.IOException; +import java.util.Collections; +import java.util.HashMap; +import java.util.Map; + +/** This class implements {@link Model} for {@link LogisticRegression}. */ +public class LogisticRegressionModel + implements Model<LogisticRegressionModel>, + LogisticRegressionModelParams<LogisticRegressionModel> { + + private Map<Param<?>, Object> paramMap = new HashMap<>(); + + private Table modelData; + + public LogisticRegressionModel() { + ParamUtils.initializeMapWithDefaultValues(this.paramMap, this); + } + + @Override + public Map<Param<?>, Object> getParamMap() { + return paramMap; + } + + @Override + public void save(String path) throws IOException { + ReadWriteUtils.saveMetadata(this, path); + ReadWriteUtils.saveModelData( + LogisticRegressionModelData.getModelDataStream(modelData), + path, + new LogisticRegressionModelData.ModelDataEncoder()); + } + + public static LogisticRegressionModel load(StreamExecutionEnvironment env, String path) + throws IOException { + LogisticRegressionModel model = ReadWriteUtils.loadStageParam(path); + Table modelData = + ReadWriteUtils.loadModelData( + env, path, new LogisticRegressionModelData.ModelDataDecoder()); + return model.setModelData(modelData); + } + + @Override + public LogisticRegressionModel setModelData(Table... inputs) { + modelData = inputs[0]; + return this; + } + + @Override + public Table[] getModelData() { + return new Table[] {modelData}; + } + + @Override + @SuppressWarnings("unchecked") + public Table[] transform(Table... inputs) { + Preconditions.checkArgument(inputs.length == 1); + StreamTableEnvironment tEnv = + (StreamTableEnvironment) ((TableImpl) inputs[0]).getTableEnvironment(); + DataStream<Row> inputStream = tEnv.toDataStream(inputs[0]); + final String broadcastModelKey = "broadcastModelKey"; + DataStream<LogisticRegressionModelData> modelData = + LogisticRegressionModelData.getModelDataStream(this.modelData); + RowTypeInfo inputTypeInfo = TableUtils.getRowTypeInfo(inputs[0].getResolvedSchema()); + RowTypeInfo outputTypeInfo = + new RowTypeInfo( + ArrayUtils.addAll( + inputTypeInfo.getFieldTypes(), + BasicTypeInfo.DOUBLE_TYPE_INFO, + TypeInformation.of(DenseVector.class)), + ArrayUtils.addAll( + inputTypeInfo.getFieldNames(), + getPredictionCol(), + getRawPredictionCol())); + DataStream<Row> predictionResult = + BroadcastUtils.withBroadcastStream( + Collections.singletonList(inputStream), + Collections.singletonMap(broadcastModelKey, modelData), + inputList -> { + DataStream inputData = inputList.get(0); + return inputData.transform( + "doPrediction", + outputTypeInfo, + new PredictOperator(broadcastModelKey, getFeaturesCol())); + }); + return new Table[] {tEnv.fromDataStream(predictionResult)}; + } + + /** A utility operator used for prediction. */ + private static class PredictOperator + extends AbstractUdfStreamOperator<Row, AbstractRichFunction> + implements OneInputStreamOperator<Row, Row> { + + private final String broadcastModelKey; + + private final String featuresCol; + + private DenseVector coefficient; + + public PredictOperator(String broadcastModelKey, String featuresCol) { + super(new AbstractRichFunction() {}); + this.broadcastModelKey = broadcastModelKey; + this.featuresCol = featuresCol; + } + + @Override + public void processElement(StreamRecord<Row> streamRecord) { + if (coefficient == null) { + LogisticRegressionModelData modelData = + (LogisticRegressionModelData) + userFunction Review comment: Hmm, I don't think this is the long term plan to support `withBroadcast`. Now `withBroadcast` is marked as internal. For the long run, we may need to discuss how to extend the existing implementations for `BroadcastState` and there may be a flip for this. -- 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
