关联规则FP树算法java实现
首先，定义一个节点：
package algorithm.correlation.fptree;
import java.util.ArrayList;
import java.util.List;
/**
* 树节点模型
* Title: UCAS <br>
* Description: <br>
* Date: 2014-6-23 <br>
* Copyright (c) 2014 AILK <br>
*
* @author fengbo
*/
public class TreeNode implements Comparable<TreeNode> {
private String name; // 节点名称
private int count; // 计数
private TreeNode parent; // 父节点
private List<TreeNode> children; // 子节点
private TreeNode nextSameNode; // 下一个同名节点
public TreeNode(String item) {
}
public TreeNode() {
}
public void countIncrement(int n) {
this.count += n;
}
public String getName() {
return name;
}
public void setName(String name) {
this.name = name;
}
public int getCount() {
return count;
}
public void setCount(int count) {
this.count = count;
}
public TreeNode getParent() {
return parent;
}
public void setParent(TreeNode parent) {
this.parent = parent;
}
public List<TreeNode> getChildren() {
return children;
}
public void setChildren(List<TreeNode> children) {
this.children = children;
}
public TreeNode getNextSameNode() {
return nextSameNode;
}
public void setNextSameNode(TreeNode nextSameNode) {
this.nextSameNode = nextSameNode;
}
public int compareTo(TreeNode treeNode) {
int count0 = treeNode.getCount();
return count0 - this.count;
}
/**
* 添加一个子节点
* @param child
* @author fengbo
*/
public void addChild(TreeNode child) {
if (this.getChildren() == null) {
List<TreeNode> list = new ArrayList<TreeNode>();
list.add(child);
this.setChildren(list);
} else {
this.getChildren().add(child);
}
}
/**
* 查找一个子节点
* @param name
* @return
* @author fengbo
*/
public TreeNode findChild(String name) {
List<TreeNode> childName = this.getChildren();
if(childName != null)
{
for (TreeNode child : childName)
{
if(name.equals(child.getName()))
{
return child;
}
}
}
return null;
}
/**
* 打印节点
*
* @author fengbo
*/
public void printChildrenName() {
List<TreeNode> children = this.getChildren();
if (children != null) {
for (TreeNode child : children) {
System.out.print(child.getName() + " ");
}
} else {
System.out.print("null");
}
}
然后，在写其他的方法：
package algorithm.correlation.fptree;
import java.util.ArrayList;
import java.util.Collections;
import java.util.Comparator;
import java.util.HashMap;
import java.util.Iterator;
import java.util.LinkedList;
import java.util.List;
import java.util.Map;
import java.util.Set;
import java.util.Map.Entry;
/**
* FP树算法实现 Title: UCAS <br>
* Description: <br>
* Date: 2014-6-23 <br>
* Copyright (c) 2014 AILK <br>
*
* @author fengbo
*/
public class FPTree {
private int minSup; // 最小支持度
// 初始化交易记录
private final static List<String> transList = new ArrayList<String>(); // 所有交易
static {
transList.add("f,a,c,d,g,i,m,p");
transList.add("a,b,c,f,l,m,o");
transList.add("b,f,h,j,o");
transList.add("b,c,k,s,p");
transList.add("a,f,c,e,l,p,m,n");
}
public int getMinSup() {
return minSup;
}
public void setMinSup(int minSup) {
this.minSup = minSup;
}
/**
* 1.读入事务记录
*
* @param filenames
* @return
* @author fengbo
*/
public List<List<String>> readTransData() {
List<List<String>> records = new LinkedList<List<String>>();
List<String> record;
for(String str : transList)
{
record = new LinkedList<String>();
String[] tran = str.split(",");
for(String t : tran)
{
record.add(t);
}
records.add(record);
}
return records;
}
/**
* 2.构造频繁1项集
*
* @param transRecords
* @return
* @author fengbo
*/
public ArrayList<TreeNode> buildF1Items(List<List<String>> transRecords) {
ArrayList<TreeNode> F1 = null;
if (transRecords.size() > 0) {
F1 = new ArrayList<TreeNode>();
Map<String, TreeNode> map = new HashMap<String, TreeNode>();
// 计算支持度
for (List<String> record : transRecords) {
for (String item : record) {
if (!map.keySet().contains(item)) {
TreeNode node = new TreeNode(item);
node.setCount(1);
map.put(item, node);
} else {
map.get(item).countIncrement(1);
}
}
}
// 把支持度大于（或等于）minSup的项加入到F1中
Set<String> names = map.keySet();
for (String name : names) {
TreeNode tnode = map.get(name);
if (tnode.getCount() >= minSup) {
tnode.setName(name);
F1.add(tnode);
}
}
Collections.sort(F1);
return F1;
} else {
return null;
}
}
/**
* 3建立FP树
*
* @param transRecords
* @param F1
* @return
* @author fengbo
*/
public TreeNode buildFPTree(List<List<String>> transRecords, ArrayList<TreeNode> F1) {
TreeNode root = new TreeNode(); // 创建树的根节点
for (List<String> transRecord : transRecords) {
LinkedList<String> record = sortByF1(transRecord, F1);
TreeNode subTreeRoot = root;
TreeNode tmpRoot = null;
if (root.getChildren() != null) {
while (!record.isEmpty()
&& (tmpRoot = subTreeRoot.findChild(record.peek())) != null) {
tmpRoot.countIncrement(1);
subTreeRoot = tmpRoot;
record.poll();
}
}
addNodes(subTreeRoot, record, F1);
}
return root;
}
/**
* 3.1把事务数据库中的一条记录按照F1（频繁1项集）中的顺序排序
*
* @param transRecord
* @param F1
* @return
* @author fengbo
*/
public LinkedList<String> sortByF1(List<String> transRecord, ArrayList<TreeNode> F1) {
Map<String, Integer> map = new HashMap<String, Integer>();
for (String item : transRecord) {
for (int i = 0; i < F1.size(); i++) {
TreeNode tnode = F1.get(i);
if (item.equals(tnode.getName())) {
map.put(item, i);
}
}
}
ArrayList<Entry<String, Integer>> al = new ArrayList<Entry<String, Integer>>(map.entrySet());
Collections.sort(al, new Comparator<Map.Entry<String, Integer>>() {
public int compare(Entry<String, Integer> arg0, Entry<String, Integer> arg1) {
return arg0.getValue() - arg1.getValue();
}
});
LinkedList<String> rest = new LinkedList<String>();
for (Entry<String, Integer> entry : al) {
rest.add(entry.getKey());
}
return rest;
}
/**
* 3.2 把若干个节点作为指定指定节点的后代插入树中
*
* @param ancestor
* @param record
* @param F1
* @author fengbo
*/
public void addNodes(TreeNode ancestor, LinkedList<String> record, ArrayList<TreeNode> F1) {
if (record.size() > 0) {
while (record.size() > 0) {
String item = record.poll();
TreeNode leafnode = new TreeNode(item);
leafnode.setName(item);
leafnode.setCount(1);
leafnode.setParent(ancestor);
ancestor.addChild(leafnode);
for (TreeNode f1 : F1) {
if (f1.getName().equals(item)) {
while (f1.getNextSameNode() != null) {
f1 = f1.getNextSameNode();
}
f1.setNextSameNode(leafnode);
break;
}
}
addNodes(leafnode, record, F1);
}
}
}
/**
* 4. 从FPTree中找到所有的频繁模式
*
* @param root
* @param F1
* @return
* @author fengbo
*/
public Map<List<String>, Integer> findFP(TreeNode root, ArrayList<TreeNode> F1) {
Map<List<String>, Integer> fp = new HashMap<List<String>, Integer>();
Iterator<TreeNode> iter = F1.iterator();
while (iter.hasNext()) {
TreeNode curr = iter.next();
// 寻找cur的条件模式基CPB，放入transRecords中
List<List<String>> transRecords = new LinkedList<List<String>>();
TreeNode backnode = curr.getNextSameNode();
while (backnode != null) {
int counter = backnode.getCount();
List<String> prenodes = new ArrayList<String>();
TreeNode parent = backnode;
// 遍历backnode的祖先节点，放到prenodes中
while ((parent = parent.getParent()).getName() != null) {
prenodes.add(parent.getName());
}
while (counter-- > 0) {
transRecords.add(prenodes);
}
backnode = backnode.getNextSameNode();
}
// 生成条件频繁1项集
ArrayList<TreeNode> subF1 = buildF1Items(transRecords);
// 建立条件模式基的局部FP-tree
TreeNode subRoot = buildFPTree(transRecords, subF1);
// 从条件FP-Tree中寻找频繁模式
if (subRoot != null) {
Map<List<String>, Integer> prePatterns = findPrePattern(subRoot);
if (prePatterns != null) {
Set<Entry<List<String>, Integer>> ss = prePatterns.entrySet();
for (Entry<List<String>, Integer> entry : ss) {
entry.getKey().add(curr.getName());
fp.put(entry.getKey(), entry.getValue());
}
}
}
}
return fp;
}
/**
* 4.1 从一棵FP-Tree上找到所有的前缀模式
*
* @param root
* @return
* @author fengbo
*/
public Map<List<String>, Integer> findPrePattern(TreeNode root) {
Map<List<String>, Integer> patterns = null;
List<TreeNode> children = root.getChildren();
if (children != null) {
patterns = new HashMap<List<String>, Integer>();
for (TreeNode child : children) {
// 找到以child为根节点的子树中的所有长路径（所谓长路径指它不是其他任何路径的子路径）
LinkedList<LinkedList<TreeNode>> paths = buildPaths(child);
if (paths != null) {
for (List<TreeNode> path : paths) {
Map<List<String>, Integer> backPatterns = combination(path);
Set<Entry<List<String>, Integer>> entryset = backPatterns.entrySet();
for (Entry<List<String>, Integer> entry : entryset) {
List<String> key = entry.getKey();
int c1 = entry.getValue();
int c0 = 0;
if (patterns.containsKey(key)) {
c0 = patterns.get(key).byteValue();
}
patterns.put(key, c0 + c1);
}
}
}
}
}
// 过滤掉那些小于MinSup的模式
Map<List<String>, Integer> rect = null;
if (patterns != null) {
rect = new HashMap<List<String>, Integer>();
Set<Entry<List<String>, Integer>> ss = patterns.entrySet();
for (Entry<List<String>, Integer> entry : ss) {
if (entry.getValue() >= minSup) {
rect.put(entry.getKey(), entry.getValue());
}
}
}
return rect;
}
/**
* 4.1.1 找到从指定节点（root）到所有可达叶子节点的路径
*
* @param root
* @return
* @author fengbo
*/
public LinkedList<LinkedList<TreeNode>> buildPaths(TreeNode root) {
LinkedList<LinkedList<TreeNode>> paths = null;
if (root != null) {
paths = new LinkedList<LinkedList<TreeNode>>();
List<TreeNode> children = root.getChildren();
if (children != null) {
// 在从树上分离单条路径时，对分叉口的节点，其count也要分到各条路径上去
// 条件FP-Tree是多枝的情况
if (children.size() > 1) {
for (TreeNode child : children) {
int count = child.getCount();
LinkedList<LinkedList<TreeNode>> ll = buildPaths(child);
for (LinkedList<TreeNode> lp : ll) {
TreeNode prenode = new TreeNode(root.getName());
prenode.setCount(count);
lp.addFirst(prenode);
paths.add(lp);
}
}
}
// 条件FP-Tree是单枝的情况
else {
for (TreeNode child : children) {
LinkedList<LinkedList<TreeNode>> ll = buildPaths(child);
for (LinkedList<TreeNode> lp : ll) {
lp.addFirst(root);
paths.add(lp);
}
}
}
} else {
LinkedList<TreeNode> lp = new LinkedList<TreeNode>();
lp.add(root);
paths.add(lp);
}
}
return paths;
}
/**
* 4.1.2 生成路径path中所有元素的任意组合，并记下每一种组合的count--其实就是组合中最后一个元素的count， 因为我们的组合算法保证了树中
*
* @param path
* @return
* @author fengbo
*/
public Map<List<String>, Integer> combination(List<TreeNode> path) {
if (path.size() > 0) {
// 从path中移除首节点
TreeNode start = path.remove(0);
// 首节点自己可以成为一个组合，放入rect中
Map<List<String>, Integer> rect = new HashMap<List<String>, Integer>();
List<String> li = new ArrayList<String>();
li.add(start.getName());
rect.put(li, start.getCount());
Map<List<String>, Integer> postCombination = combination(path);
if (postCombination != null) {
Set<Entry<List<String>, Integer>> set = postCombination.entrySet();
for (Entry<List<String>, Integer> entry : set) {
// 把首节点之后元素的所有组合放入rect中
rect.put(entry.getKey(), entry.getValue());
// 首节点并上其后元素的各种组合放入rect中
List<String> ll = new ArrayList<String>();
ll.addAll(entry.getKey());
ll.add(start.getName());
rect.put(ll, entry.getValue());
}
}
return rect;
} else {
return null;
}
}
/**
* 打印频繁1项集
*
* @param F1
* @author fengbo
*/
public void printF1(List<TreeNode> F1) {
System.out.println("F-1 set: ");
for (TreeNode item : F1) {
System.out.print(item.getName() + ":" + item.getCount() + "\t");
}
System.out.println();
System.out.println();
}
/**
* 打印FP-Tree
*
* @param root
* @author fengbo
*/
public void printFPTree(TreeNode root) {
printNode(root);
List<TreeNode> children = root.getChildren();
if (children != null && children.size() > 0) {
for (TreeNode child : children) {
printFPTree(child);
}
}
}
/**
* 打印树上单个节点的信息
*
* @param node
* @author fengbo
*/
public void printNode(TreeNode node) {
if (node.getName() != null) {
System.out.print("Name:" + node.getName() + "\tCount:" + node.getCount() + "\tParent:"
+ node.getParent().getName());
if (node.getNextSameNode() != null)
System.out.print("\tNextSameNode:" + node.getNextSameNode().getName());
System.out.print("\tChildren:");
node.printChildrenName();
System.out.println();
} else {
System.out.println("FPTreeRoot");
}
}
/**
* 打印最终找到的所有频繁模式集
*
* @param patterns
*/
public void printFreqPatterns(Map<List<String>, Integer> patterns) {
System.out.println();
System.out.println("最小支持度：" + this.getMinSup());
System.out.println("满足最小支持度的所有频繁模式集如下：");
Set<Entry<List<String>, Integer>> ss = patterns.entrySet();
for (Entry<List<String>, Integer> entry : ss) {
List<String> list = entry.getKey();
for (String item : list) {
System.out.print(item + " ");
}
System.out.print("\t" + entry.getValue());
System.out.println();
}
}
public static void main(String[] args) {
FPTree fptree = new FPTree();
fptree.setMinSup(3);
List<List<String>> transRecords = fptree.readTransData();
System.out.println(transRecords);
System.out.println("读取数据完毕！--------------------------------------------");
ArrayList<TreeNode> F1 = fptree.buildF1Items(transRecords);
fptree.printF1(F1);
System.out.println("找出频繁1项集完毕！----------------------------------------");
TreeNode treeroot = fptree.buildFPTree(transRecords, F1);
fptree.printFPTree(treeroot);
System.out.println("创建树完毕！----------------------------------------------");
Map<List<String>, Integer> patterns = fptree.findFP(treeroot, F1);
fptree.printFreqPatterns(patterns);
System.out.println("找出频繁项完毕---------------------------------------------！");
}
}
本方法包含2个类，TreeNode是定义了一个基本节点， FPTree是主要的类。该程序是可以直接运行，在借鉴了前人的经验之后，自己写的一点东西，但是笔者认为还有不足之处，在readTransData方法里的2层for循环，我认为不太好：
for(String str : transList
{
record = new LinkedList<String>();
String[] tran = str.split(",");
for(String t : tran)
{
record.add(t);
}
records.add(record);
}
这里是对元数据进行加工收集，目前我也没有想出啥好办法，细想之中，发表博文，望广大朋友可以提出自己的高见，谢谢！