ALINK(三十二):特征工程(十一)特征选择(三)向量卡方选择器 (VectorChiSqSelectorBatchOp)
Java 类名:com.alibaba.alink.operator.batch.feature.VectorChiSqSelectorBatchOp
Python 类名:VectorChiSqSelectorBatchOp
功能介绍
针对vector数据,进行特征筛选
参数说明
名称 |
中文名称 |
描述 |
类型 |
是否必须? |
默认值 |
labelCol |
标签列名 |
输入表中的标签列名 |
String |
✓ |
|
selectedCol |
选中的列名 |
计算列对应的列名 |
String |
✓ |
|
selectorType |
筛选类型 |
筛选类型,包含"NumTopFeatures","percentile", "fpr", "fdr", "fwe"五种。 |
String |
"NumTopFeatures" |
|
numTopFeatures |
最大的p-value列个数 |
最大的p-value列个数, 默认值50 |
Integer |
50 |
|
percentile |
筛选的百分比 |
筛选的百分比,默认值0.1 |
Double |
0.1 |
|
fpr |
p value的阈值 |
p value的阈值,默认值0.05 |
Double |
0.05 |
|
fdr |
发现阈值 |
发现阈值, 默认值0.05 |
Double |
0.05 |
|
fwe |
错误率阈值 |
错误率阈值, 默认值0.05 |
Double |
0.05 |
代码示例
Python 代码
无python接口
Java 代码
package javatest.com.alibaba.alink.batch.feature; import org.apache.flink.types.Row; import com.alibaba.alink.operator.batch.BatchOperator; import com.alibaba.alink.operator.batch.feature.VectorChiSqSelectorBatchOp; import com.alibaba.alink.operator.batch.source.MemSourceBatchOp; import org.junit.Test; import java.util.Arrays; public class VectorChiSqSelectorBatchOpTest { @Test public void testVectorChiSqSelectorBatchOp() throws Exception { Row[] testArray = new Row[] { Row.of(7, "0.0 0.0 18.0 1.0", 1.0), Row.of(8, "0.0 1.0 12.0 0.0", 0.0), Row.of(9, "1.0 0.0 15.0 0.1", 0.0), }; String[] colNames = new String[] {"id", "features", "clicked"}; MemSourceBatchOp source = new MemSourceBatchOp(Arrays.asList(testArray), colNames); VectorChiSqSelectorBatchOp test = new VectorChiSqSelectorBatchOp() .setSelectedCol("features") .setLabelCol("clicked"); test.linkFrom(source); test.lazyPrintModelInfo(); BatchOperator.execute(); } }
运行结果
------------------------- ChisqSelectorModelInfo ------------------------- Number of Selector Features: 4 Number of Features: 4 Type of Selector: NumTopFeatures Number of Top Features: 50 Selector Indices: |VectorIndex|ChiSquare|PValue| DF|Selected| |-----------|---------|------|---|--------| | 3| 3|0.2231| 2| true| | 2| 3|0.2231| 2| true| | 0| 0.75|0.3865| 1| true| | 1| 0.75|0.3865| 1| true|
本文来自博客园,作者:秋华,转载请注明原文链接:https://www.cnblogs.com/qiu-hua/p/14901601.html