zhipeng93 commented on a change in pull request #73: URL: https://github.com/apache/flink-ml/pull/73#discussion_r834144685
########## File path: flink-ml-lib/src/main/java/org/apache/flink/ml/feature/standardscaler/StandardScalerModel.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.feature.standardscaler; + +import org.apache.flink.api.common.functions.RichMapFunction; +import org.apache.flink.api.common.typeinfo.TypeInformation; +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.SparseVector; +import org.apache.flink.ml.linalg.Vector; +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.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; + +/** A Model which transforms data using the model data computed by {@link StandardScaler}. */ +public class StandardScalerModel + implements Model<StandardScalerModel>, StandardScalerParams<StandardScalerModel> { + private final Map<Param<?>, Object> paramMap = new HashMap<>(); + private Table modelDataTable; + + public StandardScalerModel() { + ParamUtils.initializeMapWithDefaultValues(paramMap, this); + } + + @Override + @SuppressWarnings("unchecked, rawtypes") + public Table[] transform(Table... inputs) { + Preconditions.checkArgument(inputs.length == 1); + StreamTableEnvironment tEnv = + (StreamTableEnvironment) ((TableImpl) inputs[0]).getTableEnvironment(); + DataStream<Row> inputStream = tEnv.toDataStream(inputs[0]); + + RowTypeInfo inputTypeInfo = TableUtils.getRowTypeInfo(inputs[0].getResolvedSchema()); + RowTypeInfo outputTypeInfo = + new RowTypeInfo( + ArrayUtils.addAll( + inputTypeInfo.getFieldTypes(), TypeInformation.of(Vector.class)), + ArrayUtils.addAll(inputTypeInfo.getFieldNames(), getPredictionCol())); + + final String broadcastModelKey = "broadcastModelKey"; + DataStream<StandardScalerModelData> modelDataStream = + StandardScalerModelData.getModelDataStream(modelDataTable); + + DataStream<Row> predictionResult = + BroadcastUtils.withBroadcastStream( + Collections.singletonList(inputStream), + Collections.singletonMap(broadcastModelKey, modelDataStream), + inputList -> { + DataStream inputData = inputList.get(0); + return inputData.map( + new PredictOutputFunction( + broadcastModelKey, + getFeaturesCol(), + getWithMean(), + getWithStd()), + outputTypeInfo); + }); + + return new Table[] {tEnv.fromDataStream(predictionResult)}; + } + + /** A utility function used for prediction. */ + private static class PredictOutputFunction extends RichMapFunction<Row, Row> { + private final String broadcastModelKey; + private final String featuresCol; + private final boolean withMean; + private final boolean withStd; + private DenseVector mean; + private DenseVector scale; + + public PredictOutputFunction( + String broadcastModelKey, String featuresCol, boolean withMean, boolean withStd) { + this.broadcastModelKey = broadcastModelKey; + this.featuresCol = featuresCol; + this.withMean = withMean; + this.withStd = withStd; + } + + @Override + public Row map(Row dataPoint) { + if (mean == null) { + StandardScalerModelData modelData = + (StandardScalerModelData) + getRuntimeContext().getBroadcastVariable(broadcastModelKey).get(0); + mean = modelData.mean; + DenseVector std = modelData.std; + + if (withStd) { + scale = std; + double[] scaleValues = scale.values; + for (int i = 0; i < scaleValues.length; i++) { + scaleValues[i] = scaleValues[i] == 0 ? 0 : 1 / scaleValues[i]; + } + } + } + + Vector feature = ((Vector) dataPoint.getField(featuresCol)).clone(); + + if (withMean && feature instanceof SparseVector) { + feature = Vectors.toDense((SparseVector) feature); Review comment: `toDense` is not used for performance here... It is because we can only support in-place adding a `SparseVector/DenseVector` to a `DenseVecor` in `axpy`. -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
