[GitHub] [flink-ml] zhipeng93 commented on a diff in pull request #83: [FLINK-27170] Add Transformer and Estimator for OnlineLogisticRegression

[GitBox <gi...@apache.org>]

zhipeng93 commented on code in PR #83:
URL: https://github.com/apache/flink-ml/pull/83#discussion_r886644705


##########
flink-ml-lib/src/main/java/org/apache/flink/ml/classification/logisticregression/OnlineLogisticRegression.java:
##########
@@ -0,0 +1,434 @@
+/*
+ * 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.logisticregression;
+
+import org.apache.flink.api.common.functions.FilterFunction;
+import org.apache.flink.api.common.functions.MapFunction;
+import org.apache.flink.api.common.functions.ReduceFunction;
+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.typeutils.ObjectArrayTypeInfo;
+import org.apache.flink.iteration.DataStreamList;
+import org.apache.flink.iteration.IterationBody;
+import org.apache.flink.iteration.IterationBodyResult;
+import org.apache.flink.iteration.Iterations;
+import org.apache.flink.iteration.operator.OperatorStateUtils;
+import org.apache.flink.ml.api.Estimator;
+import org.apache.flink.ml.common.datastream.DataStreamUtils;
+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.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.streaming.api.datastream.DataStream;
+import org.apache.flink.streaming.api.operators.AbstractStreamOperator;
+import org.apache.flink.streaming.api.operators.OneInputStreamOperator;
+import org.apache.flink.streaming.api.operators.TwoInputStreamOperator;
+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.collections.IteratorUtils;
+
+import java.io.IOException;
+import java.util.Arrays;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+
+/**
+ * An Estimator which implements the FTRL-Proximal online learning algorithm proposed by H. Brendan
+ * McMahan et al.
+ *
+ * <p>See <a href="https://doi.org/10.1145/2487575.2488200">H. Brendan McMahan et al., Ad click
+ * prediction: a view from the trenches.</a>
+ */
+public class OnlineLogisticRegression
+        implements Estimator<OnlineLogisticRegression, OnlineLogisticRegressionModel>,
+                OnlineLogisticRegressionParams<OnlineLogisticRegression> {
+    private final Map<Param<?>, Object> paramMap = new HashMap<>();
+    private Table initModelDataTable;
+
+    public OnlineLogisticRegression() {
+        ParamUtils.initializeMapWithDefaultValues(paramMap, this);
+    }
+
+    @Override
+    @SuppressWarnings("unchecked")
+    public OnlineLogisticRegressionModel fit(Table... inputs) {
+        Preconditions.checkArgument(inputs.length == 1);
+
+        StreamTableEnvironment tEnv =
+                (StreamTableEnvironment) ((TableImpl) inputs[0]).getTableEnvironment();
+        DataStream<LogisticRegressionModelData> modelDataStream =
+                LogisticRegressionModelData.getModelDataStream(initModelDataTable);
+
+        DataStream<Row> points =
+                tEnv.toDataStream(inputs[0])
+                        .map(new FeaturesExtractor(getFeaturesCol(), getLabelCol()));
+
+        DataStream<DenseVector> initModelData =
+                modelDataStream.map(
+                        (MapFunction<LogisticRegressionModelData, DenseVector>)
+                                value -> value.coefficient);

Review Comment:
   If the init model version is 5, then will you create a model with version smaller than 5?



##########
flink-ml-lib/src/main/java/org/apache/flink/ml/classification/logisticregression/LogisticRegressionModelData.java:
##########
@@ -18,40 +18,86 @@
 
 package org.apache.flink.ml.classification.logisticregression;
 
+import org.apache.flink.api.common.functions.MapFunction;
 import org.apache.flink.api.common.serialization.Encoder;
 import org.apache.flink.api.common.typeinfo.TypeInformation;
 import org.apache.flink.configuration.Configuration;
 import org.apache.flink.connector.file.src.reader.SimpleStreamFormat;
 import org.apache.flink.core.fs.FSDataInputStream;
 import org.apache.flink.core.memory.DataInputViewStreamWrapper;
 import org.apache.flink.core.memory.DataOutputViewStreamWrapper;
+import org.apache.flink.ml.common.datastream.TableUtils;
 import org.apache.flink.ml.linalg.DenseVector;
 import org.apache.flink.ml.linalg.typeinfo.DenseVectorSerializer;
 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 java.io.EOFException;
 import java.io.IOException;
 import java.io.OutputStream;
+import java.util.Random;
 
 /**
- * Model data of {@link LogisticRegressionModel}.
+ * Model data of {@link LogisticRegressionModel} and {@link OnlineLogisticRegressionModel}.
  *
  * <p>This class also provides methods to convert model data from Table to Datastream, and classes
  * to save/load model data.
  */
 public class LogisticRegressionModelData {
 
     public DenseVector coefficient;
+    public long modelVersion;
 
-    public LogisticRegressionModelData(DenseVector coefficient) {
+    /* todo : modelVersion is a new idea to manage model, other algorithm if needed should add this parameter later. */
+    public LogisticRegressionModelData(DenseVector coefficient, long modelVersion) {
         this.coefficient = coefficient;
+        this.modelVersion = modelVersion;
     }
 
     public LogisticRegressionModelData() {}
 
+    public LogisticRegressionModelData(DenseVector coefficient) {
+        this.coefficient = coefficient;
+        this.modelVersion = 0L;
+    }
+
+    /**
+     * Generates a Table containing a {@link LogisticRegressionModelData} instance with randomly
+     * generated coefficient.
+     *
+     * @param tEnv The environment where to create the table.
+     * @param dim The size of generated coefficient.
+     * @param seed Random seed.
+     */
+    public static Table generateRandomModelData(StreamTableEnvironment tEnv, int dim, int seed) {
+        StreamExecutionEnvironment env = TableUtils.getExecutionEnvironment(tEnv);
+        return tEnv.fromDataStream(env.fromElements(1).map(new GenerateRandomModel(dim, seed)));
+    }
+
+    private static class GenerateRandomModel

Review Comment:
   This comment is marked as resolved but seems not addressed. Did you forget about this?



##########
flink-ml-core/src/test/java/org/apache/flink/ml/common/datastream/DataStreamUtilsTest.java:
##########
@@ -74,6 +75,15 @@ public void testReduce() throws Exception {
         assertArrayEquals(new long[] {190L}, sum.stream().mapToLong(Long::longValue).toArray());
     }
 
+    @Test
+    public void testGenerateBatchData() throws Exception {
+        DataStream<Long> dataStream =
+                env.fromParallelCollection(new NumberSequenceIterator(0L, 19L), Types.LONG);
+        DataStream<Long[]> result = DataStreamUtils.generateBatchData(dataStream, 2, 4);
+        List<Long[]> batches = IteratorUtils.toList(result.executeAndCollect());
+        assertEquals(10, batches.size());

Review Comment:
   Could you please check the size of each local batch?



##########
flink-ml-core/src/main/java/org/apache/flink/ml/common/datastream/DataStreamUtils.java:
##########
@@ -182,4 +193,63 @@ public void snapshotState(StateSnapshotContext context) throws Exception {
             }
         }
     }
+
+    /**
+     * A function that generate several data batches and distribute them to downstream operator.
+     *
+     * @param <T> Data type of batch data.
+     */
+    public static <T> DataStream<T[]> generateBatchData(

Review Comment:
   This comment is not addressed nut marked as resolved. Did you forget to address it?



-- 
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