均值算法的c语言程序,k均值算法c++语言实现代码

#include "k-mean.h"

#include

#include

#include

#include

#include

#include

#include

using namespace std;

#define MAXDISTANCE 1000000

KMean::KMean(int c_num,vector node_vector)

{

cluster_num = c_num;

data = node_vector;

srand((int)time(0));

Init();

}

void KMean::Init()//初始化函数(首先随即生成代表点)

{

for(int i =0 ;i

{

int pos = rand() % data.size();

bool insert_flag = true;

//首先判断选中的点是否是中心点

for(unsigned int j = 0;j

{

if(mean_nodes[j] ==  data[i])

{

insert_flag = false;

break;

}

}

if(insert_flag )

{

mean_nodes.push_back(data[pos]);

i++;

}

}

ClusterProcess();//进行聚类过程

}

void KMean::ClusterProcess()//聚类过程，将空间中的点分到不同的簇中

{

//遍历空间上所有的点

for( unsigned int i = 0 ; i < data.size();i++)

{

//忽略中心点

bool continue_flag = false;

for(unsigned int j = 0;j

{

if(mean_nodes[j] ==  data[i])

{

continue_flag = true;

break;

}

}

if(continue_flag)

{

continue;

}

//下面是聚类过程

//首先找到离当前点最近的中心点,并记录下该中心点所在的簇

int min_kdistance = MAXDISTANCE;

int index = 0 ;

for(unsigned int j = 0;j < mean_nodes.size();j++)

{

double dis = Kdistance(data[i],mean_nodes[j]);

if(dis < min_kdistance)

{

min_kdistance = dis;

index = j;

}

}

//先将当前点从原先的簇中删除

map >::iterator iter;

//搜索所有的簇

for(iter = cluster.begin();iter != cluster.end();++iter)

{

vector::iterator node_iter;

bool jump_flag = false;

//对每个簇中的vector进行搜索

for(node_iter = iter->second.begin();node_iter != iter->second.end();node_iter++)

{

if(*node_iter == data[i])

{

//如果当前点就在更新的簇中，则忽略后面的操作

if(index == iter->first)

{

continue_flag = true;

}

else

{

iter->second.erase(node_iter);

}

jump_flag = true;

break;

}

}

if(jump_flag)

{

break;

}

}

if(continue_flag)

{

continue;

}

//将当前点插入到中心点所对应的簇中

//查看中心点是否已经存在map中

bool insert_flag = true;

for(iter = cluster.begin(); iter != cluster.end();++iter)

{

if(iter->first == index)

{

iter->second.push_back(data[i]);

insert_flag = false;

break;

}

}

//不存在则创建新的元素对象

if(insert_flag)

{

vector cluster_node_vector;

cluster_node_vector.push_back(data[i]);

cluster.insert(make_pair(index,cluster_node_vector));

}

}

}

double KMean::Kdistance(Node active,Node other)

{

return sqrt(pow((active.pos_x-other.pos_x),2) + pow((active.pos_y - other.pos_y),2) + pow((active.pos_z - other.pos_z),2));

}

Node KMean::GetMean(int cluster_index)

{

//对传入的参数进行判断，查看是否越界

if( cluster_num <0 || unsigned (cluster_index) >= mean_nodes.size() )

{

Node new_node;

new_node.pos_x = -1.0;

new_node.pos_y = -1.0;

new_node.pos_z = -1.0;

return new_node;

}

//求出簇中所有点的均值

Node sum_node;

Node aver_node;

for(int j = 0;j < cluster[cluster_index].size();j++)

{

sum_node.pos_x += cluster[cluster_index].at(j).pos_x;

sum_node.pos_y += cluster[cluster_index].at(j).pos_y;

sum_node.pos_z += cluster[cluster_index].at(j).pos_z;

}

aver_node.pos_x = sum_node.pos_x*1.0/ cluster[cluster_index].size();

aver_node.pos_y = sum_node.pos_y*1.0 / cluster[cluster_index].size();

aver_node.pos_z = sum_node.pos_z*1.0 / cluster[cluster_index].size();

//找到与均值最近的点

double min_dis = MAXDISTANCE;

Node new_mean_doc;

for(unsigned int i  = 0;i< cluster[cluster_index].size();i++)

{

double dis = Kdistance(aver_node,cluster[cluster_index].at(i));

if(min_dis > dis)

{

min_dis = dis;

new_mean_doc = cluster[cluster_index].at(i);

}

}

return new_mean_doc;

}

void KMean::NewCluster()//确定新的中心点

{

for (unsigned int i = 0;i < mean_nodes.size();i++)

{

Node new_node =GetMean(i);

mean_nodes[i] = new_node;

}

ClusterProcess();

}

void KMean::Star()

{

for (int i = 0;i<100;i++)

{

NewCluster();

cout << "no:"<< i<

for(int j = 0;j < mean_nodes.size();j++)

{

cout << cluster[j].size()<

}

}

}