/**
* ���ýڵ����
*
* @param clusteringFeature ��ǰ�ڵ�
* @param level ��ǰ���ֵ
*/
private void setTreeLevel(ClusteringFeature clusteringFeature, int level) {
LeafNode leafNode = null;
NonLeafNode nonLeafNode = null;
if (clusteringFeature instanceof LeafNode) {
leafNode = (LeafNode) clusteringFeature;
} else if (clusteringFeature instanceof NonLeafNode) {
nonLeafNode = (NonLeafNode) clusteringFeature;
}
if (nonLeafNode != null) {
nonLeafNode.setLevel(level);
level++;
// �����ӽڵ�
if (nonLeafNode.getNonLeafChilds() != null) {
for (NonLeafNode n1 : nonLeafNode.getNonLeafChilds()) {
setTreeLevel(n1, level);
}
} else {
for (LeafNode n2 : nonLeafNode.getLeafChilds()) {
setTreeLevel(n2, level);
}
}
} else {
leafNode.setLevel(level);
level++;
// �����Ӿ۴�
for (Cluster c : leafNode.getClusterChilds()) {
c.setLevel(level);
}
}
}
/**
* ����CF����������
*
* @return
*/
private ClusteringFeature buildCFTree() {
NonLeafNode rootNode = null;
LeafNode leafNode = null;
Cluster cluster = null;
for (String[] record : totalDataRecords) {
cluster = new Cluster(record);
if (rootNode == null) {
// CF��ֻ��1���ڵ��ʱ������
if (leafNode == null) {
leafNode = new LeafNode();
}
leafNode.addingCluster(cluster);
if (leafNode.getParentNode() != null) {
rootNode = leafNode.getParentNode();
}
} else {
if (rootNode.getParentNode() != null) {
rootNode = rootNode.getParentNode();
}
// �Ӹ�ڵ㿪ʼ����������Ѱ�ҵ��������Ŀ��Ҷ�ӽڵ�
LeafNode temp = rootNode.findedClosestNode(cluster);
temp.addingCluster(cluster);
}
}
// ���������ҳ�������Ľڵ�
LeafNode node = cluster.getParentNode();
NonLeafNode upNode = node.getParentNode();
if (upNode == null) {
return node;
} else {
while (upNode.getParentNode() != null) {
upNode = upNode.getParentNode();
}
return upNode;
}
}
public void add(double x, double y, int depth, LeafNode<T> value) {
if ((depth % 2) == 0) {
// current node is split on X axis
if (x < splitValue) {
// left subtree
if (left == null) {
left = value;
} else {
if (left.isLeaf()) {
// left == leaf node , split at Y-axis and re-insert
final LeafNode<T> tmp = (LeafNode<T>) left;
if (tmp.x == x && tmp.y == y) {
if (!tmp.supportsMultipleValues()) {
left = new MultiValuedLeafNode<>((SingleValueLeafNode<T>) tmp);
}
left.add(x, y, depth, value);
} else {
final double splitY = (tmp.y + y) / 2.0;
final NonLeafNode<T> newNode = new NonLeafNode<>(splitY);
newNode.add(tmp.x, tmp.y, depth + 1, tmp);
newNode.add(x, y, depth + 1, value);
left = newNode;
}
} else {
left.add(x, y, depth + 1, value);
}
}
} else {
// right subtree
if (right == null) {
right = value;
} else {
if (right.isLeaf()) {
// right == leaf node , split at Y-axis and re-insert
final LeafNode<T> tmp = (LeafNode<T>) right;
if (tmp.x == x && tmp.y == y) {
if (!tmp.supportsMultipleValues()) {
right = new MultiValuedLeafNode<>((SingleValueLeafNode<T>) tmp);
}
right.add(x, y, depth, value);
} else {
final double splitY = (tmp.y + y) / 2.0;
final NonLeafNode<T> newNode = new NonLeafNode<>(splitY);
newNode.add(tmp.x, tmp.y, depth + 1, tmp);
newNode.add(x, y, depth + 1, value);
right = newNode;
}
} else {
right.add(x, y, depth + 1, value);
}
}
}
} else {
// current node is split on Y axis
if (y < splitValue) {
// left subtree
if (left == null) {
left = value;
} else {
if (left.isLeaf()) {
// left == leaf node , split at X-axis and re-insert
final LeafNode<T> tmp = (LeafNode<T>) left;
if (tmp.x == x && tmp.y == y) {
if (!tmp.supportsMultipleValues()) {
left = new MultiValuedLeafNode<>((SingleValueLeafNode<T>) tmp);
}
left.add(x, y, depth, value);
} else {
final double splitX = (tmp.x + x) / 2.0;
final NonLeafNode<T> newNode = new NonLeafNode<>(splitX);
newNode.add(tmp.x, tmp.y, depth + 1, tmp);
newNode.add(x, y, depth + 1, value);
left = newNode;
}
} else {
left.add(x, y, depth + 1, value);
}
}
} else {
// right subtree
if (right == null) {
right = value;
} else {
if (right.isLeaf()) {
// right == leaf node , split at X-axis and re-insert
final LeafNode<T> tmp = (LeafNode<T>) right;
if (tmp.x == x && tmp.y == y) {
if (!tmp.supportsMultipleValues()) {
right = new MultiValuedLeafNode<>((SingleValueLeafNode<T>) tmp);
}
right.add(x, y, depth, value);
} else {
final double splitX = (tmp.x + x) / 2.0;
final NonLeafNode<T> newNode = new NonLeafNode<>(splitX);
newNode.add(tmp.x, tmp.y, depth + 1, tmp);
newNode.add(x, y, depth + 1, value);
right = newNode;
}
} else {
right.add(x, y, depth + 1, value);
}
}
}
}
}
/**
* ��ʾCF����������
*
* @param rootNode CF����ڵ�
*/
private void showCFTree(ClusteringFeature rootNode) {
// �ո����������
int blankNum = 5;
// ��ǰ�����
int currentLevel = 1;
LinkedList<ClusteringFeature> nodeQueue = new LinkedList<>();
ClusteringFeature cf;
LeafNode leafNode;
NonLeafNode nonLeafNode;
ArrayList<Cluster> clusterList = new ArrayList<>();
String typeName;
nodeQueue.add(rootNode);
while (nodeQueue.size() > 0) {
cf = nodeQueue.poll();
if (cf instanceof LeafNode) {
leafNode = (LeafNode) cf;
typeName = LEAFNODE;
if (leafNode.getClusterChilds() != null) {
for (Cluster c : leafNode.getClusterChilds()) {
nodeQueue.add(c);
}
}
} else if (cf instanceof NonLeafNode) {
nonLeafNode = (NonLeafNode) cf;
typeName = NON_LEAFNODE;
if (nonLeafNode.getNonLeafChilds() != null) {
for (NonLeafNode n1 : nonLeafNode.getNonLeafChilds()) {
nodeQueue.add(n1);
}
} else {
for (LeafNode n2 : nonLeafNode.getLeafChilds()) {
nodeQueue.add(n2);
}
}
} else {
clusterList.add((Cluster) cf);
typeName = CLUSTER;
}
if (currentLevel != cf.getLevel()) {
currentLevel = cf.getLevel();
System.out.println();
System.out.println("|");
System.out.println("|");
} else if (currentLevel == cf.getLevel() && currentLevel != 1) {
for (int i = 0; i < blankNum; i++) {
System.out.print("-");
}
}
System.out.print(typeName);
System.out.print("N:" + cf.getN() + ", LS:");
System.out.print("[");
for (double d : cf.getLS()) {
System.out.print(MessageFormat.format("{0}, ", d));
}
System.out.print("]");
}
System.out.println();
System.out.println("*******���շֺõľ۴�****");
// ��ʾ�Ѿ��ֺ���ľ۴ص�
for (int i = 0; i < clusterList.size(); i++) {
System.out.println("Cluster" + (i + 1) + "��");
for (double[] point : clusterList.get(i).getData()) {
System.out.print("[");
for (double d : point) {
System.out.print(MessageFormat.format("{0}, ", d));
}
System.out.println("]");
}
}
}
// 节点分裂
@Override
public void split() {
int c1 = 0;
int c2 = 0;
double maxDist = 0;
int length = this.getChildrenList().size();
for (int i = 0; i < length - 1; i++) {
for (int j = i + 1; j < length; j++) {
double dist =
this.getChildrenList()
.get(i)
.getClusterFeature()
.getDistanceTo(this.getChildrenList().get(j).getClusterFeature());
if (dist > maxDist) {
maxDist = dist;
c1 = i;
c2 = j;
}
}
}
// 以距离最远的孩子节点为中心，把B+1个孩子分为两个大簇，其中一个簇仍留做本节点的孩子，另一个新建一个节点来领养它们
LeafNode newNode = new LeafNode();
newNode.addChild(this.getChildrenList().get(c2));
// 如果本节点已经是root节点，则需要创建一个新的root节点
if (this.getParent() == null) {
NonLeafNode root = new NonLeafNode();
root.setN(this.getN());
root.setLS(this.getLS());
root.setSS(this.getSS());
this.setParent(root);
root.addChild(this);
}
// 建立新节点和本节点的父节点的父子关系
newNode.setParent(this.getParent());
((NonLeafNode) this.getParent()).addChild(newNode);
// 把离newNode节点近的孩子归属到这个newNode簇里面
for (int i = 0; i < length; i++) {
if (i != c1 && i != c2) {
if (this.getChildrenList()
.get(i)
.getClusterFeature()
.getDistanceTo(this.getChildrenList().get(c2).getClusterFeature())
< this.getChildrenList()
.get(i)
.getClusterFeature()
.getDistanceTo(this.getChildrenList().get(c1).getClusterFeature())) {
newNode.addChild(this.getChildrenList().get(i));
}
}
}
// 把离newNode近的孩子从本节点中删除
for (MinCluster cluster : newNode.getChildrenList()) {
newNode.addCF(cluster.getClusterFeature(), true);
this.deleteChild(cluster);
this.addCF(cluster.getClusterFeature(), false);
}
// 把新增加的LeafNode添加到LeafNode的双向链表中
if (this.getNext() != null) {
newNode.setNext(this.getNext());
this.getNext().setPre(newNode);
}
this.setNext(newNode);
newNode.setPre(this);
// 如果本节点的分裂导致了父节点的孩子书超过了分支因子，则引发父节点的分裂
NonLeafNode pn = (NonLeafNode) this.getParent();
if (pn.getChildrenList().size() > pn.getB()) {
this.getParent().split();
}
}