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 2022/07/29 09:01:01 UTC
[GitHub] [flink-ml] zhipeng93 commented on a diff in pull request #133: [FLINK-28601] Add Transformer for FeatureHasher
zhipeng93 commented on code in PR #133:
URL: https://github.com/apache/flink-ml/pull/133#discussion_r933001756
##########
flink-ml-lib/src/main/java/org/apache/flink/ml/feature/featurehasher/FeatureHasher.java:
##########
@@ -0,0 +1,199 @@
+/*
+ * 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.featurehasher;
+
+import org.apache.flink.api.common.functions.MapFunction;
+import org.apache.flink.api.java.typeutils.RowTypeInfo;
+import org.apache.flink.ml.api.Transformer;
+import org.apache.flink.ml.common.datastream.TableUtils;
+import org.apache.flink.ml.linalg.SparseVector;
+import org.apache.flink.ml.linalg.typeinfo.VectorTypeInfo;
+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.table.api.DataTypes;
+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.catalog.ResolvedSchema;
+import org.apache.flink.table.types.DataType;
+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.ArrayList;
+import java.util.Arrays;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.TreeMap;
+
+import static org.apache.flink.shaded.guava30.com.google.common.hash.Hashing.murmur3_32;
+
+/**
+ * FeatureHasher is a transformer that projects a set of categorical or numerical features into a
+ * sparse vector. This is done using the hashing trick
+ * (https://en.wikipedia.org/wiki/Feature_hashing) to map features to indices in the feature vector.
+ */
+public class FeatureHasher
+ implements Transformer<FeatureHasher>, FeatureHasherParams<FeatureHasher> {
+ private final Map<Param<?>, Object> paramMap = new HashMap<>();
+ private static final org.apache.flink.shaded.guava30.com.google.common.hash.HashFunction HASH =
+ murmur3_32(0);
+
+ public FeatureHasher() {
+ ParamUtils.initializeMapWithDefaultValues(paramMap, this);
+ }
+
+ @Override
+ public Table[] transform(Table... inputs) {
+ Preconditions.checkArgument(inputs.length == 1);
+ StreamTableEnvironment tEnv =
+ (StreamTableEnvironment) ((TableImpl) inputs[0]).getTableEnvironment();
+ ResolvedSchema tableSchema = inputs[0].getResolvedSchema();
+ RowTypeInfo inputTypeInfo = TableUtils.getRowTypeInfo(tableSchema);
+ RowTypeInfo outputTypeInfo =
+ new RowTypeInfo(
+ ArrayUtils.addAll(inputTypeInfo.getFieldTypes(), VectorTypeInfo.INSTANCE),
+ ArrayUtils.addAll(inputTypeInfo.getFieldNames(), getOutputCol()));
+ DataStream<Row> output =
+ tEnv.toDataStream(inputs[0])
+ .map(
+ new HashFunction(
+ getInputCols(),
+ generateCategoricalCols(
+ tableSchema, getInputCols(), getCategoricalCols()),
+ getNumFeatures()),
+ outputTypeInfo);
+ Table outputTable = tEnv.fromDataStream(output);
+ return new Table[] {outputTable};
+ }
+
+ /**
Review Comment:
How about update the java doc as:
`
The main logic for transforming the categorical and numerical features into a sparse vector. It uses MurMurHash3 to compute the transformed index in the output vector. If multiple features are projected to the same column, their values are accumulated.
`
##########
flink-ml-lib/src/main/java/org/apache/flink/ml/feature/featurehasher/FeatureHasher.java:
##########
@@ -0,0 +1,199 @@
+/*
+ * 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.featurehasher;
+
+import org.apache.flink.api.common.functions.MapFunction;
+import org.apache.flink.api.java.typeutils.RowTypeInfo;
+import org.apache.flink.ml.api.Transformer;
+import org.apache.flink.ml.common.datastream.TableUtils;
+import org.apache.flink.ml.linalg.SparseVector;
+import org.apache.flink.ml.linalg.typeinfo.VectorTypeInfo;
+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.table.api.DataTypes;
+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.catalog.ResolvedSchema;
+import org.apache.flink.table.types.DataType;
+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.ArrayList;
+import java.util.Arrays;
+import java.util.HashMap;
+import java.util.List;
+import java.util.Map;
+import java.util.TreeMap;
+
+import static org.apache.flink.shaded.guava30.com.google.common.hash.Hashing.murmur3_32;
+
+/**
Review Comment:
nit: How about we update the Java doc as well as the python doc as follows?
```
A Transformer that transforms a set of categorical or numerical features into a sparse vector of a specified dimension. The rules of hashing categorical columns and numerical columns are as follows:
<ul>
<li> For numerical columns, the index of this feature in the output vector is the hash value of the column name and its correponding value is the same as the input.
<li>For categorical columns, the index of this feature in the output vector is the hash value of the string "column_name=value" and the corresponding value is 1.0.
</ul>
<p>If multiple features are projected into the same column, the output values are accumulated. For the hashing trick, see https://en.wikipedia.org/wiki/Feature_hashing for details.
```
##########
flink-ml-python/pyflink/examples/ml/feature/vectorassembler_example.py:
##########
@@ -58,7 +58,7 @@
.set_output_col('assembled_vec') \
.set_handle_invalid('keep')
-# use the vector assembler model for feature engineering
+# use the vector assembler object for feature engineering
Review Comment:
how about simply remove `object` here?
--
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