IForest异常检测 (IForestOutlierStreamOp)

Java 类名：com.alibaba.alink.operator.stream.outlier.IForestOutlierStreamOp

Python 类名：IForestOutlierStreamOp

功能介绍

iForest 可以识别数据中异常点，在异常检测领域有比较好的效果。算法使用 sub-sampling 方法，降低了算法的计算复杂度。

文献或出处

1. Isolation Forest

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
predictionCol	预测结果列名	预测结果列名	String	✓
featureCols	特征列名数组	特征列名数组，默认全选	String[]		所选列类型为 [BIGDECIMAL, BIGINTEGER, BYTE, DOUBLE, FLOAT, INTEGER, LONG, SHORT]	null
groupCols	分组列名数组	分组列名，多列，可选，默认不选	String[]			null
numTrees	模型中树的棵数	模型中树的棵数	Integer			100
outlierThreshold	异常评分阈值	只有评分大于该阈值才会被认为是异常点	Double
precedingRows	数据窗口大小	数据窗口大小	Integer			null
precedingTime	时间窗口大小	时间窗口大小	String			null
predictionDetailCol	预测详细信息列名	预测详细信息列名	String
subsamplingSize	每棵树的样本采样行数	每棵树的样本采样行数，默认 256 ，最小 2 ，最大 100000 .	Integer		1 <= x <= 100000	256
tensorCol	tensor列	tensor列	String		所选列类型为 [BOOL_TENSOR, BYTE_TENSOR, DOUBLE_TENSOR, FLOAT_TENSOR, INT_TENSOR, LONG_TENSOR, STRING, STRING_TENSOR, TENSOR, UBYTE_TENSOR]	null
timeCol	时间戳列(TimeStamp)	时间戳列(TimeStamp)	String			null
vectorCol	向量列名	向量列对应的列名，默认值是null	String		所选列类型为 [DENSE_VECTOR, SPARSE_VECTOR, STRING, VECTOR]	null
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

import time, datetime
import numpy as np
import pandas as pd

data = pd.DataFrame([
			[1, datetime.datetime.fromtimestamp(1), 10.0, 0],
			[1, datetime.datetime.fromtimestamp(2), 11.0, 0],
			[1, datetime.datetime.fromtimestamp(3), 12.0, 0],
			[1, datetime.datetime.fromtimestamp(4), 13.0, 0],
			[1, datetime.datetime.fromtimestamp(5), 14.0, 0],
			[1, datetime.datetime.fromtimestamp(6), 15.0, 0],
			[1, datetime.datetime.fromtimestamp(7), 16.0, 0],
			[1, datetime.datetime.fromtimestamp(8), 17.0, 0],
			[1, datetime.datetime.fromtimestamp(9), 18.0, 0],
			[1, datetime.datetime.fromtimestamp(10), 19.0, 0]
])

dataOp = dataframeToOperator(data, schemaStr='id int, ts timestamp, val double, label int', op_type='stream')

outlierOp = IForestOutlierStreamOp()\
			.setGroupCols(["id"])\
			.setTimeCol("ts")\
			.setPrecedingRows(3)\
			.setFeatureCols(["val"])\
			.setPredictionCol("pred")\
			.setPredictionDetailCol("pred_detail")

dataOp.link(outlierOp).print()

StreamOperator.execute()

Java 代码

package com.alibaba.alink.operator.stream.outlier;

import org.apache.flink.types.Row;

import com.alibaba.alink.operator.stream.StreamOperator;
import com.alibaba.alink.operator.stream.source.MemSourceStreamOp;
import com.alibaba.alink.testutil.AlinkTestBase;
import org.junit.Test;

import java.sql.Timestamp;
import java.util.Arrays;
import java.util.List;

public class IForestOutlierStreamOpTest extends AlinkTestBase {

	@Test
	public void test() throws Exception {
		List <Row> mTableData = Arrays.asList(
			Row.of(1, new Timestamp(1), 10.0, 0),
			Row.of(1, new Timestamp(2), 11.0, 0),
			Row.of(1, new Timestamp(3), 12.0, 0),
			Row.of(1, new Timestamp(4), 13.0, 0),
			Row.of(1, new Timestamp(5), 14.0, 0),
			Row.of(1, new Timestamp(6), 15.0, 0),
			Row.of(1, new Timestamp(7), 16.0, 0),
			Row.of(1, new Timestamp(8), 17.0, 0),
			Row.of(1, new Timestamp(9), 18.0, 0),
			Row.of(1, new Timestamp(10), 19.0, 0)
		);

		MemSourceStreamOp dataOp = new MemSourceStreamOp(mTableData, new String[] {"id", "ts", "val", "label"});

		IForestOutlierStreamOp outlierOp = new IForestOutlierStreamOp()
			.setGroupCols("id")
			.setTimeCol("ts")
			.setPrecedingRows(3)
			.setFeatureCols("val")
			.setPredictionCol("pred")
			.setPredictionDetailCol("pred_detail");

		dataOp.link(outlierOp).print();

		StreamOperator.execute();

	}
}

运行结果

无