zhipeng93 commented on a change in pull request #24: URL: https://github.com/apache/flink-ml/pull/24#discussion_r768484621
########## File path: flink-ml-lib/src/main/java/org/apache/flink/ml/classification/knn/Knn.java ########## @@ -0,0 +1,155 @@ +/* + * 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.knn; + +import org.apache.flink.api.common.functions.MapFunction; +import org.apache.flink.api.common.functions.RichMapPartitionFunction; +import org.apache.flink.api.java.tuple.Tuple3; +import org.apache.flink.ml.api.Estimator; +import org.apache.flink.ml.common.datastream.DataStreamUtils; +import org.apache.flink.ml.linalg.DenseMatrix; +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.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.Collector; +import org.apache.flink.util.Preconditions; + +import java.io.IOException; +import java.util.ArrayList; +import java.util.HashMap; +import java.util.List; +import java.util.Map; + +/** + * An Estimator which implements the KNN algorithm. + * + * <p>See: https://en.wikipedia.org/wiki/K-nearest_neighbors_algorithm. + */ +public class Knn implements Estimator<Knn, KnnModel>, KnnParams<Knn> { + + private final Map<Param<?>, Object> paramMap = new HashMap<>(); + + public Knn() { + ParamUtils.initializeMapWithDefaultValues(paramMap, this); + } + + @Override + public KnnModel fit(Table... inputs) { + Preconditions.checkArgument(inputs.length == 1); + StreamTableEnvironment tEnv = + (StreamTableEnvironment) ((TableImpl) inputs[0]).getTableEnvironment(); + /* Tuple3 : <feature, label, norm> */ + DataStream<Tuple3<DenseVector, Double, Double>> inputDataWithNorm = + computeNorm(tEnv.toDataStream(inputs[0])); + DataStream<KnnModelData> modelData = genModelData(inputDataWithNorm); + KnnModel model = new KnnModel().setModelData(tEnv.fromDataStream(modelData)); + ReadWriteUtils.updateExistingParams(model, getParamMap()); + return model; + } + + @Override + public Map<Param<?>, Object> getParamMap() { + return paramMap; + } + + @Override + public void save(String path) throws IOException { + ReadWriteUtils.saveMetadata(this, path); + } + + public static Knn load(StreamExecutionEnvironment env, String path) throws IOException { + return ReadWriteUtils.loadStageParam(path); + } + + /** + * Generates knn model data. For Euclidean distance, distance = sqrt((a - b)^2) = (sqrt(a^2 + + * b^2 - 2ab)) So it can pre-calculate the L2 norm square of the feature vector, and when + * calculating the distance with another feature vector, only dot product is calculated. + * + * @param inputDataWithNorm Input data with feature norm. + * @return Knn model. + */ + private static DataStream<KnnModelData> genModelData( + DataStream<Tuple3<DenseVector, Double, Double>> inputDataWithNorm) { + DataStream<KnnModelData> modelData = + DataStreamUtils.mapPartition( + inputDataWithNorm, + new RichMapPartitionFunction< + Tuple3<DenseVector, Double, Double>, KnnModelData>() { + @Override + public void mapPartition( + Iterable<Tuple3<DenseVector, Double, Double>> values, + Collector<KnnModelData> out) { + List<Tuple3<DenseVector, Double, Double>> buffer = + new ArrayList<>(); + for (Tuple3<DenseVector, Double, Double> value : values) { + buffer.add(value); + } + int featureDim = buffer.get(0).f0.size(); + DenseMatrix packedFeatures = + new DenseMatrix(featureDim, buffer.size()); + DenseVector featureNorms = new DenseVector(buffer.size()); + DenseVector labels = new DenseVector(buffer.size()); + int offset = 0; + for (Tuple3<DenseVector, Double, Double> data : buffer) { Review comment: data -> dataPoint -- 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]
