[GitHub] [flink-ml] zhipeng93 commented on a diff in pull request #132: [FLINK-28571] Add AlgoOperator for Chi-squared test

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

zhipeng93 commented on code in PR #132:
URL: https://github.com/apache/flink-ml/pull/132#discussion_r932805286


##########
flink-ml-lib/src/main/java/org/apache/flink/ml/stats/chisqtest/ChiSqTest.java:
##########
@@ -0,0 +1,492 @@
+/*
+ * 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.stats.chisqtest;
+
+import org.apache.flink.api.common.functions.MapPartitionFunction;
+import org.apache.flink.api.common.functions.RichFlatMapFunction;
+import org.apache.flink.api.common.functions.RichMapFunction;
+import org.apache.flink.api.common.functions.RichMapPartitionFunction;
+import org.apache.flink.api.common.typeinfo.TypeInformation;
+import org.apache.flink.api.common.typeinfo.Types;
+import org.apache.flink.api.java.tuple.Tuple2;
+import org.apache.flink.api.java.tuple.Tuple3;
+import org.apache.flink.api.java.tuple.Tuple4;
+import org.apache.flink.api.java.typeutils.RowTypeInfo;
+import org.apache.flink.ml.api.AlgoOperator;
+import org.apache.flink.ml.common.broadcast.BroadcastUtils;
+import org.apache.flink.ml.common.datastream.DataStreamUtils;
+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.datastream.SingleOutputStreamOperator;
+import org.apache.flink.streaming.api.operators.AbstractStreamOperator;
+import org.apache.flink.streaming.api.operators.BoundedOneInput;
+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.Collector;
+import org.apache.flink.util.Preconditions;
+
+import org.apache.commons.math3.distribution.ChiSquaredDistribution;
+
+import java.io.IOException;
+import java.math.BigDecimal;
+import java.math.RoundingMode;
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.List;
+import java.util.Map;
+import java.util.Optional;
+import java.util.function.Function;
+import java.util.stream.Collectors;
+
+/**
+ * Chi-square test of independence of variables in a contingency table. This Transformer computes
+ * the chi-square statistic,p-value,and dof(number of degrees of freedom) for every feature in the
+ * contingency table, which constructed from the `observed` for each categorical values. All label
+ * and feature values must be categorical.
+ *
+ * <p>See: http://en.wikipedia.org/wiki/Chi-squared_test.
+ */
+public class ChiSqTest implements AlgoOperator<ChiSqTest>, ChiSqTestParams<ChiSqTest> {
+
+    final String bcCategoricalMarginsKey = "bcCategoricalMarginsKey";
+    final String bcLabelMarginsKey = "bcLabelMarginsKey";
+
+    private final Map<Param<?>, Object> paramMap = new HashMap<>();
+
+    public ChiSqTest() {
+        ParamUtils.initializeMapWithDefaultValues(paramMap, this);
+    }
+
+    @Override
+    public Table[] transform(Table... inputs) {
+
+        final String[] inputCols = getInputCols();
+        String labelCol = getLabelCol();
+
+        StreamTableEnvironment tEnv =
+                (StreamTableEnvironment) ((TableImpl) inputs[0]).getTableEnvironment();
+
+        SingleOutputStreamOperator<Tuple3<String, Object, Object>> colAndFeatureAndLabel =
+                tEnv.toDataStream(inputs[0])
+                        .flatMap(new ExtractColAndFeatureAndLabel(inputCols, labelCol));
+
+        // compute the observed frequencies
+        DataStream<Tuple4<String, Object, Object, Long>> observedFreq =
+                DataStreamUtils.mapPartition(
+                        colAndFeatureAndLabel.keyBy(Tuple3::hashCode),
+                        new GenerateObservedFrequencies());
+
+        SingleOutputStreamOperator<Tuple4<String, Object, Object, Long>> filledObservedFreq =

Review Comment:
   Thanks for the explanation.
   
   For 1: I agree that `GenerateObservedFrequencies` will reduce most of the input data in most cases, but there exist cases that there are many `zeros` in one column, e.g, there are many different labels but data is sparse. 
   However, postponing the `FillZeroFunc` should always outperforms the current implementation.
   
   For 2, I think we will not output `FilledObservedFreq` in `ChiSqTest`. We should be careful about the behavior of a well-known algorithm --- `ChiSqTest` does  not output the contingency table. If we need to output the contingency table, we may have a new algorithm for that.
   
   



-- 
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