二值化 (Binarizer)

Java 类名：com.alibaba.alink.pipeline.feature.Binarizer

Python 类名：Binarizer

功能介绍

给定一个阈值，将连续变量二值化（大于等于阈值转为1，小于阈值转为0）。

参数说明

名称	中文名称	描述	类型	是否必须？	取值范围	默认值
selectedCol	选中的列名	计算列对应的列名	String	✓
outputCol	输出结果列	输出结果列列名，可选，默认null	String			null
reservedCols	算法保留列名	算法保留列	String[]			null
threshold	二值化阈值	二值化阈值	Double			0.0
numThreads	组件多线程线程个数	组件多线程线程个数	Integer			1

代码示例

Python 代码

from pyalink.alink import *

import pandas as pd

useLocalEnv(1)

df = pd.DataFrame([
    [1.1, True, "2", "A"],
    [1.1, False, "2", "B"],
    [1.1, True, "1", "B"],
    [2.2, True, "1", "A"]
])

inOp = BatchOperator.fromDataframe(df, schemaStr='double double, bool boolean, number int, str string')
binarizer = Binarizer().setSelectedCol("double").setThreshold(2.0)
binarizer.transform(inOp).print()

Java 代码

import org.apache.flink.types.Row;

import com.alibaba.alink.operator.batch.BatchOperator;
import com.alibaba.alink.operator.batch.source.MemSourceBatchOp;
import com.alibaba.alink.pipeline.feature.Binarizer;
import org.junit.Test;

import java.util.Arrays;
import java.util.List;

public class BinarizerTest {
	@Test
	public void testBinarizer() throws Exception {
		List <Row> df = Arrays.asList(
			Row.of(1.1, true, 2, "A"),
			Row.of(1.1, false, 2, "B"),
			Row.of(1.1, true, 1, "B"),
			Row.of(2.2, true, 1, "A")
		);
		BatchOperator <?> inOp = new MemSourceBatchOp(df, "double double, bool boolean, number int, str string");
		Binarizer binarizer = new Binarizer().setSelectedCol("double").setThreshold(2.0);
		binarizer.transform(inOp).print();
	}
}

运行结果

double	bool	number	str
0.0000	true	2	A
0.0000	false	2	B
0.0000	true	1	B
1.0000	true	1	A