分组FPGrowth训练 (GroupedFpGrowthBatchOp)

Java 类名：com.alibaba.alink.operator.batch.associationrule.GroupedFpGrowthBatchOp

Python 类名：GroupedFpGrowthBatchOp

功能介绍

分组FpGrowth组件按照指定的分组列，在每个分组内使用FpGrowth算法进行频繁项集挖掘。

FpGrowth算法详见FpGrowthBatchOp

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
itemsCol	项集列名	项集列名	String	✓	所选列类型为 [STRING]
groupCol	分组单列名	分组单列名，可选	String			null
maxConsequentLength	最大关联规则后继长度	最大关联规则后继(consequent)长度	Integer			1
maxPatternLength	最大频繁项集长度	最大频繁项集长度	Integer			10
minConfidence	最小置信度	最小置信度，同时包含X和Y的样本与包含X的样本之比，反映了当样本中包含项集X时，项集Y同时出现的概率。	Double			0.05
minLift	最小提升度	最小提升度，提升度是用来衡量A出现的情况下，是否会对B出现的概率有所提升。	Double			1.0
minSupportCount	最小支持度数目	最小支持度目，当取值大于或等于0时起作用，当小于0时参数minSupportPercent起作用	Integer			-1
minSupportPercent	最小支持度占比	最小支持度占比，当minSupportCount取值小于0时起作用，当minSupportCount大于或等于0时该参数不起作用	Double			0.02

<!–

代码示例

Python 代码

from pyalink.alink import *

import pandas as pd

useLocalEnv(1)

df = pd.DataFrame([
    ["changjiang", "A,B,C,D"],
    ["changjiang", "B,C,E"],
    ["huanghe", "A,B,C,E"],
    ["huanghe", "B,D,E"],
    ["huanghe", "A,B,C,D"],
])

data = BatchOperator.fromDataframe(df, schemaStr='group string, items string')

fpGrowth = GroupedFpGrowthBatchOp() \
    .setItemsCol("items").setGroupCol("group").setMinSupportCount(2)

fpGrowth.linkFrom(data)

fpGrowth.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 org.junit.Test;

import java.util.Arrays;

public class GroupedFpGrowthBatchOpTest {
	@Test
	public void test() throws Exception {
		Row[] rows = new Row[] {
			Row.of("changjiang", "A,B,C,D"),
			Row.of("changjiang", "B,C,E"),
			Row.of("huanghe", "A,B,C,E"),
			Row.of("huanghe", "B,D,E"),
			Row.of("huanghe", "A,B,C,D"),
		};

		BatchOperator data = new MemSourceBatchOp(Arrays.asList(rows), "group string, items string");

		BatchOperator fpgrowth = new GroupedFpGrowthBatchOp()
			.setGroupCol("group")
			.setItemsCol("items")
			.setMinSupportCount(2);

		fpgrowth.linkFrom(data);
		fpgrowth.print();
	}
}

三元组输入

import org.apache.flink.types.Row;

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

import java.util.Arrays;

public class GroupedFpGrowthBatchOpTest {
	@Test
	public void test() throws Exception {
		Row[] rows = new Row[] {
			Row.of("changjiang", 1, "A"),
			Row.of("changjiang", 1, "B"),
			Row.of("changjiang", 1, "C"),
			Row.of("changjiang", 1, "D"),
			Row.of("changjiang", 2, "B"),
			Row.of("changjiang", 2, "C"),
			Row.of("changjiang", 2, "E"),
			Row.of("huanghe", 3, "A"),
			Row.of("huanghe", 3, "B"),
			Row.of("huanghe", 3, "C"),
			Row.of("huanghe", 3, "E"),
			Row.of("huanghe", 4, "B"),
			Row.of("huanghe", 4, "D"),
			Row.of("huanghe", 4, "E"),
			Row.of("huanghe", 5, "A"),
			Row.of("huanghe", 5, "B"),
			Row.of("huanghe", 5, "C"),
			Row.of("huanghe", 5, "D"),
		};

		BatchOperator data = new MemSourceBatchOp(Arrays.asList(rows), "groupid string, id int, item string")
			.groupBy("groupid,id", "groupid,id,CONCAT_AGG(item) AS items");

		BatchOperator fpgrowth = new GroupedFpGrowthBatchOp()
			.setGroupCol("groupid")
			.setItemsCol("items")
			.setMinSupportCount(2);

		fpgrowth.linkFrom(data);
		fpgrowth.print();
	}
}

运行结果

group	itemset	supportcount	itemcount
changjiang	B	2	1
changjiang	C	2	1
changjiang	B,C	2	2
huanghe	A	2	1
huanghe	B,A	2	2
huanghe	B	3	1
huanghe	C	2	1
huanghe	B,C	2	2
huanghe	A,C	2	2
huanghe	B,A,C	2	3
huanghe	D	2	1
huanghe	B,D	2	2
huanghe	E	2	1
huanghe	B,E	2	2