You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by GitBox <gi...@apache.org> on 2022/03/22 06:24:47 UTC

[GitHub] [flink-ml] lindong28 commented on a change in pull request #54: [FLINK-25552] Add Estimator and Transformer for MinMaxScaler

lindong28 commented on a change in pull request #54:
URL: https://github.com/apache/flink-ml/pull/54#discussion_r831805455



##########
File path: flink-ml-lib/src/test/java/org/apache/flink/ml/feature/MinMaxScalerTest.java
##########
@@ -0,0 +1,218 @@
+/*
+ * 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;
+
+import org.apache.flink.api.common.functions.MapFunction;
+import org.apache.flink.api.common.restartstrategy.RestartStrategies;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.ml.feature.minmaxscaler.MinMaxScaler;
+import org.apache.flink.ml.feature.minmaxscaler.MinMaxScalerModel;
+import org.apache.flink.ml.linalg.DenseVector;
+import org.apache.flink.ml.linalg.Vectors;
+import org.apache.flink.ml.util.ReadWriteUtils;
+import org.apache.flink.ml.util.StageTestUtils;
+import org.apache.flink.streaming.api.datastream.DataStream;
+import org.apache.flink.streaming.api.environment.ExecutionCheckpointingOptions;
+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.commons.collections.IteratorUtils;
+import org.junit.Before;
+import org.junit.Rule;
+import org.junit.Test;
+import org.junit.rules.TemporaryFolder;
+
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Collections;
+import java.util.List;
+
+import static org.junit.Assert.assertEquals;
+
+/** Tests {@link MinMaxScaler} and {@link MinMaxScalerModel}. */
+public class MinMaxScalerTest {
+    @Rule public final TemporaryFolder tempFolder = new TemporaryFolder();
+    private StreamExecutionEnvironment env;
+    private StreamTableEnvironment tEnv;
+    private Table trainDataTable;
+    private Table predictDataTable;
+    private static final List<Row> trainData =
+            new ArrayList<>(
+                    Arrays.asList(
+                            Row.of(Vectors.dense(0.0, 3.0)),
+                            Row.of(Vectors.dense(2.1, 0.0)),
+                            Row.of(Vectors.dense(4.1, 5.1)),
+                            Row.of(Vectors.dense(6.1, 8.1)),
+                            Row.of(Vectors.dense(200, 400))));
+    private static final List<Row> predictData =
+            new ArrayList<>(
+                    Arrays.asList(
+                            Row.of(Vectors.dense(150.0, 90.0)),
+                            Row.of(Vectors.dense(50.0, 40.0)),
+                            Row.of(Vectors.dense(100.0, 50.0))));
+
+    private static final List<DenseVector> expectedData =
+            new ArrayList<>(
+                    Arrays.asList(
+                            Vectors.dense(0.25, 0.1),
+                            Vectors.dense(0.5, 0.125),
+                            Vectors.dense(0.75, 0.225)));
+
+    @Before
+    public void before() {
+        Configuration config = new Configuration();
+        config.set(ExecutionCheckpointingOptions.ENABLE_CHECKPOINTS_AFTER_TASKS_FINISH, true);
+        env = StreamExecutionEnvironment.getExecutionEnvironment(config);
+        env.setParallelism(4);
+        env.enableCheckpointing(100);
+        env.setRestartStrategy(RestartStrategies.noRestart());
+        tEnv = StreamTableEnvironment.create(env);
+        trainDataTable = tEnv.fromDataStream(env.fromCollection(trainData)).as("features");
+        predictDataTable = tEnv.fromDataStream(env.fromCollection(predictData)).as("features");
+    }
+
+    private static void verifyPredictionResult(
+            Table output, String outputCol, List<DenseVector> expected) throws Exception {
+        StreamTableEnvironment tEnv =
+                (StreamTableEnvironment) ((TableImpl) output).getTableEnvironment();
+        DataStream<DenseVector> stream =
+                tEnv.toDataStream(output)
+                        .map(
+                                (MapFunction<Row, DenseVector>)
+                                        row -> (DenseVector) row.getField(outputCol));
+        List<DenseVector> result = IteratorUtils.toList(stream.executeAndCollect());
+
+        result.sort(
+                (first, second) -> {

Review comment:
       The Javadoc of `Comparator` says the following:
   
   ```
        * It is generally the case, but <i>not</i> strictly required that
        * <tt>(compare(x, y)==0) == (x.equals(y))</tt>.  Generally speaking,
        * any comparator that violates this condition should clearly indicate
        * this fact.  The recommended language is "Note: this comparator
        * imposes orderings that are inconsistent with equals."
   ```
   
   It might be better to take the vector size into consideration so that the comparator returns 0 iff `first == second`. The main benefit is that the code is self-contained and could be copied/used in other places.
   
   Alternatively we can just add the comment suggested by the comparator Javadoc.

##########
File path: flink-ml-lib/src/main/java/org/apache/flink/ml/feature/minmaxscaler/MinMaxScalerModel.java
##########
@@ -0,0 +1,179 @@
+/*
+ * 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.minmaxscaler;
+
+import org.apache.flink.api.common.functions.RichMapFunction;
+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.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.table.runtime.typeutils.ExternalTypeInfo;
+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 do a minMax scaler operation using the model data computed by {@link MinMaxScaler}.
+ */
+public class MinMaxScalerModel
+        implements Model<MinMaxScalerModel>, MinMaxScalerParams<MinMaxScalerModel> {
+    private final Map<Param<?>, Object> paramMap = new HashMap<>();
+    private Table modelDataTable;
+
+    public MinMaxScalerModel() {
+        ParamUtils.initializeMapWithDefaultValues(paramMap, this);
+    }
+
+    @Override
+    public MinMaxScalerModel setModelData(Table... inputs) {
+        modelDataTable = inputs[0];
+        return this;
+    }
+
+    @Override
+    public Table[] getModelData() {
+        return new Table[] {modelDataTable};
+    }
+
+    @Override
+    @SuppressWarnings("unchecked")
+    public Table[] transform(Table... inputs) {
+        Preconditions.checkArgument(inputs.length == 1);
+        StreamTableEnvironment tEnv =
+                (StreamTableEnvironment) ((TableImpl) inputs[0]).getTableEnvironment();
+        DataStream<Row> data = tEnv.toDataStream(inputs[0]);
+        DataStream<MinMaxScalerModelData> minMaxScalerModel =
+                MinMaxScalerModelData.getModelDataStream(modelDataTable);
+        final String broadcastModelKey = "broadcastModelKey";
+        RowTypeInfo inputTypeInfo = TableUtils.getRowTypeInfo(inputs[0].getResolvedSchema());
+        RowTypeInfo outputTypeInfo =
+                new RowTypeInfo(
+                        ArrayUtils.addAll(
+                                inputTypeInfo.getFieldTypes(),
+                                ExternalTypeInfo.of(DenseVector.class)),
+                        ArrayUtils.addAll(inputTypeInfo.getFieldNames(), getOutputCol()));
+        DataStream<Row> output =
+                BroadcastUtils.withBroadcastStream(
+                        Collections.singletonList(data),
+                        Collections.singletonMap(broadcastModelKey, minMaxScalerModel),
+                        inputList -> {
+                            DataStream input = inputList.get(0);
+                            return input.map(
+                                    new PredictLabelFunction(
+                                            broadcastModelKey,
+                                            getMax(),
+                                            getMIN(),
+                                            getFeaturesCol()),
+                                    outputTypeInfo);
+                        });
+        return new Table[] {tEnv.fromDataStream(output)};
+    }
+
+    @Override
+    public Map<Param<?>, Object> getParamMap() {
+        return paramMap;
+    }
+
+    @Override
+    public void save(String path) throws IOException {
+        ReadWriteUtils.saveMetadata(this, path);
+        ReadWriteUtils.saveModelData(
+                MinMaxScalerModelData.getModelDataStream(modelDataTable),
+                path,
+                new MinMaxScalerModelData.ModelDataEncoder());
+    }
+
+    /**
+     * Loads model data from path.
+     *
+     * @param env Stream execution environment.
+     * @param path Model path.
+     * @return MinMaxScalerModel model.
+     */
+    public static MinMaxScalerModel load(StreamExecutionEnvironment env, String path)
+            throws IOException {
+        StreamTableEnvironment tEnv = StreamTableEnvironment.create(env);
+        MinMaxScalerModel model = ReadWriteUtils.loadStageParam(path);
+        DataStream<MinMaxScalerModelData> modelData =
+                ReadWriteUtils.loadModelData(
+                        env, path, new MinMaxScalerModelData.ModelDataDecoder());
+        return model.setModelData(tEnv.fromDataStream(modelData));
+    }
+
+    /** This operator loads model data and predicts result. */
+    private static class PredictLabelFunction extends RichMapFunction<Row, Row> {
+        private final String featureCol;
+        private MinMaxScalerModelData minMaxScalerModelData;
+        private final double max;
+        private final double min;
+        private final String broadcastKey;
+        private DenseVector maxVector;
+        private DenseVector minVector;
+
+        public PredictLabelFunction(
+                String broadcastKey, double max, double min, String featureCol) {
+            this.max = max;
+            this.min = min;
+            this.broadcastKey = broadcastKey;
+            this.featureCol = featureCol;
+        }
+
+        @Override
+        public Row map(Row row) {
+            if (minMaxScalerModelData == null) {
+                minMaxScalerModelData =
+                        (MinMaxScalerModelData)
+                                getRuntimeContext().getBroadcastVariable(broadcastKey).get(0);
+                maxVector = minMaxScalerModelData.maxVector;
+                minVector = minMaxScalerModelData.minVector;
+            }
+            DenseVector feature = (DenseVector) row.getField(featureCol);
+            DenseVector outputVector = new DenseVector(maxVector.size());
+            if (feature != null) {

Review comment:
       @zhipeng93 @weibozhao Currently in all other model classes, we don't explicitly check whether the feature value is null. Should we follow the same practice here for consistency?
   
   We could alternatively choose to consistently check whether feature value is null, and throw exception if it is null. But even if we don't check this, we will already have NullPointerException with proper stacktrace, which should be sufficient for us to locate the problem (i.e. feature value is null). I personally think it is simpler not to check this. What do you think?
   
   
   

##########
File path: flink-ml-lib/src/main/java/org/apache/flink/ml/feature/minmaxscaler/MinMaxScalerModel.java
##########
@@ -0,0 +1,187 @@
+/*
+ * 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.minmaxscaler;
+
+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.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.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 do a minMax scaler operation using the model data computed by {@link MinMaxScaler}.
+ */
+public class MinMaxScalerModel
+        implements Model<MinMaxScalerModel>, MinMaxScalerParams<MinMaxScalerModel> {
+    private final Map<Param<?>, Object> paramMap = new HashMap<>();
+    private Table modelDataTable;
+
+    public MinMaxScalerModel() {
+        ParamUtils.initializeMapWithDefaultValues(paramMap, this);
+    }
+
+    @Override
+    public MinMaxScalerModel setModelData(Table... inputs) {
+        modelDataTable = inputs[0];
+        return this;
+    }
+
+    @Override
+    public Table[] getModelData() {
+        return new Table[] {modelDataTable};
+    }
+
+    @Override
+    @SuppressWarnings("unchecked")
+    public Table[] transform(Table... inputs) {
+        Preconditions.checkArgument(inputs.length == 1);
+        StreamTableEnvironment tEnv =
+                (StreamTableEnvironment) ((TableImpl) inputs[0]).getTableEnvironment();
+        DataStream<Row> data = tEnv.toDataStream(inputs[0]);
+        DataStream<MinMaxScalerModelData> minMaxScalerModel =
+                MinMaxScalerModelData.getModelDataStream(modelDataTable);
+        final String broadcastModelKey = "broadcastModelKey";
+        RowTypeInfo inputTypeInfo = TableUtils.getRowTypeInfo(inputs[0].getResolvedSchema());
+        RowTypeInfo outputTypeInfo =
+                new RowTypeInfo(
+                        ArrayUtils.addAll(
+                                inputTypeInfo.getFieldTypes(),
+                                TypeInformation.of(DenseVector.class)),
+                        ArrayUtils.addAll(inputTypeInfo.getFieldNames(), getPredictionCol()));
+        DataStream<Row> output =
+                BroadcastUtils.withBroadcastStream(
+                        Collections.singletonList(data),
+                        Collections.singletonMap(broadcastModelKey, minMaxScalerModel),
+                        inputList -> {
+                            DataStream input = inputList.get(0);
+                            return input.map(
+                                    new PredictOutputFunction(
+                                            broadcastModelKey,
+                                            getMax(),
+                                            getMin(),
+                                            getFeaturesCol()),
+                                    outputTypeInfo);
+                        });
+        return new Table[] {tEnv.fromDataStream(output)};
+    }
+
+    @Override
+    public Map<Param<?>, Object> getParamMap() {
+        return paramMap;
+    }
+
+    @Override
+    public void save(String path) throws IOException {
+        ReadWriteUtils.saveMetadata(this, path);
+        ReadWriteUtils.saveModelData(
+                MinMaxScalerModelData.getModelDataStream(modelDataTable),
+                path,
+                new MinMaxScalerModelData.ModelDataEncoder());
+    }
+
+    /**
+     * Loads model data from path.
+     *
+     * @param env Stream execution environment.
+     * @param path Model path.
+     * @return MinMaxScalerModel model.
+     */
+    public static MinMaxScalerModel load(StreamExecutionEnvironment env, String path)
+            throws IOException {
+        StreamTableEnvironment tEnv = StreamTableEnvironment.create(env);
+        MinMaxScalerModel model = ReadWriteUtils.loadStageParam(path);
+        DataStream<MinMaxScalerModelData> modelData =
+                ReadWriteUtils.loadModelData(
+                        env, path, new MinMaxScalerModelData.ModelDataDecoder());
+        return model.setModelData(tEnv.fromDataStream(modelData));
+    }
+
+    /** This operator loads model data and predicts result. */
+    private static class PredictOutputFunction extends RichMapFunction<Row, Row> {
+        private final String featureCol;
+        private final String broadcastKey;
+        private final double upperBound;
+        private final double lowerBound;
+        private DenseVector scaleVector;
+        private DenseVector offsetVector;
+
+        public PredictOutputFunction(
+                String broadcastKey, double upperBound, double lowerBound, String featureCol) {
+            this.upperBound = upperBound;
+            this.lowerBound = lowerBound;
+            this.broadcastKey = broadcastKey;
+            this.featureCol = featureCol;
+        }
+
+        @Override
+        public Row map(Row row) {
+            if (scaleVector == null) {
+                MinMaxScalerModelData minMaxScalerModelData =
+                        (MinMaxScalerModelData)
+                                getRuntimeContext().getBroadcastVariable(broadcastKey).get(0);
+                DenseVector minVector = minMaxScalerModelData.minVector;
+                DenseVector maxVector = minMaxScalerModelData.maxVector;
+                scaleVector = new DenseVector(minVector.size());
+                offsetVector = new DenseVector(minVector.size());
+                for (int i = 0; i < maxVector.size(); ++i) {
+                    if ((minVector.values[i] - maxVector.values[i]) == 0.0) {

Review comment:
       According to the discussion in [1][2], it is could lead to unexpected bug if we use `==` to check whether two double values are equal.
   
   How about we follow the recommendation in [3] and use the following code to check equality? The idea here is to explicitly specify an epsilon similar to what we did in tests.
   
   ```
   Math.abs(minVector.values[i] - maxVector.values[i]) < 1e-5
   ```
   
   [1] https://stackoverflow.com/questions/25160375/comparing-double-values-for-equality-in-java
   [2] https://publicobject.com/2009/11/floating-point-equality.html
   [3] https://www.baeldung.com/java-comparing-doubles

##########
File path: flink-ml-lib/src/test/java/org/apache/flink/ml/feature/MinMaxScalerTest.java
##########
@@ -0,0 +1,218 @@
+/*
+ * 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;
+
+import org.apache.flink.api.common.functions.MapFunction;
+import org.apache.flink.api.common.restartstrategy.RestartStrategies;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.ml.feature.minmaxscaler.MinMaxScaler;
+import org.apache.flink.ml.feature.minmaxscaler.MinMaxScalerModel;
+import org.apache.flink.ml.linalg.DenseVector;
+import org.apache.flink.ml.linalg.Vectors;
+import org.apache.flink.ml.util.ReadWriteUtils;
+import org.apache.flink.ml.util.StageTestUtils;
+import org.apache.flink.streaming.api.datastream.DataStream;
+import org.apache.flink.streaming.api.environment.ExecutionCheckpointingOptions;
+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.commons.collections.IteratorUtils;
+import org.junit.Before;
+import org.junit.Rule;
+import org.junit.Test;
+import org.junit.rules.TemporaryFolder;
+
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Collections;
+import java.util.List;
+
+import static org.junit.Assert.assertEquals;
+
+/** Tests {@link MinMaxScaler} and {@link MinMaxScalerModel}. */
+public class MinMaxScalerTest {
+    @Rule public final TemporaryFolder tempFolder = new TemporaryFolder();
+    private StreamExecutionEnvironment env;
+    private StreamTableEnvironment tEnv;
+    private Table trainDataTable;
+    private Table predictDataTable;
+    private static final List<Row> trainData =
+            new ArrayList<>(
+                    Arrays.asList(
+                            Row.of(Vectors.dense(0.0, 3.0)),
+                            Row.of(Vectors.dense(2.1, 0.0)),
+                            Row.of(Vectors.dense(4.1, 5.1)),
+                            Row.of(Vectors.dense(6.1, 8.1)),
+                            Row.of(Vectors.dense(200, 400))));
+    private static final List<Row> predictData =
+            new ArrayList<>(
+                    Arrays.asList(
+                            Row.of(Vectors.dense(150.0, 90.0)),
+                            Row.of(Vectors.dense(50.0, 40.0)),
+                            Row.of(Vectors.dense(100.0, 50.0))));
+
+    private static final List<DenseVector> expectedData =
+            new ArrayList<>(
+                    Arrays.asList(
+                            Vectors.dense(0.25, 0.1),
+                            Vectors.dense(0.5, 0.125),
+                            Vectors.dense(0.75, 0.225)));
+
+    @Before
+    public void before() {
+        Configuration config = new Configuration();
+        config.set(ExecutionCheckpointingOptions.ENABLE_CHECKPOINTS_AFTER_TASKS_FINISH, true);
+        env = StreamExecutionEnvironment.getExecutionEnvironment(config);
+        env.setParallelism(4);
+        env.enableCheckpointing(100);
+        env.setRestartStrategy(RestartStrategies.noRestart());
+        tEnv = StreamTableEnvironment.create(env);
+        trainDataTable = tEnv.fromDataStream(env.fromCollection(trainData)).as("features");
+        predictDataTable = tEnv.fromDataStream(env.fromCollection(predictData)).as("features");
+    }
+
+    private static void verifyPredictionResult(
+            Table output, String outputCol, List<DenseVector> expected) throws Exception {
+        StreamTableEnvironment tEnv =
+                (StreamTableEnvironment) ((TableImpl) output).getTableEnvironment();
+        DataStream<DenseVector> stream =
+                tEnv.toDataStream(output)
+                        .map(
+                                (MapFunction<Row, DenseVector>)
+                                        row -> (DenseVector) row.getField(outputCol));
+        List<DenseVector> result = IteratorUtils.toList(stream.executeAndCollect());
+
+        result.sort(
+                (first, second) -> {
+                    for (int x = 0; x < first.size(); x++) {
+                        int cmp = Double.compare(first.get(x), second.get(x));
+                        if (cmp != 0) {
+                            return cmp;
+                        }
+                    }
+                    return 0;
+                });
+        assertEquals(expected.size(), result.size());

Review comment:
       Do we still need this assertion given that we already have `assertEquals(expected, result)`?




-- 
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-unsubscribe@flink.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org