kmean.cpp

kmean算法的c++实现

#include "kmean.h"

void Kmean::initClusters()
{
	for(int i = 0; i < this->patterns.numOfCluster; i++)
	{
		Kmean::Cluster cluster;
		cluster.setCenter(this->patterns.getPattern(i));
		cluster.addPattern(i);
		this->clusters.push_back(cluster);
	}
}

void Kmean::classifyPattern()
{
	for(int i = 0; i < this->patterns.numOfCluster; i++)
	{
		this->clusters[i].clear();
	}
	for(int i = 0; i < this->patterns.numOfPatterns; i++)
	{
		int indexOfCluster = this->getClosestCluster(i);
		this->clusters[indexOfCluster].addPattern(i);
	}	
}

int Kmean::getClosestCluster(int i)
{
	double dist = 9.9e+99;
	int clustId = -1;
	Pattern pattern = this->patterns.getPattern(i);
	for(int j = 0; j < this->patterns.numOfCluster; j++)
	{
		double d = this->patterns.caluDistance(this->clusters[j].getCenter(),pattern);
		if(d < dist)
		{
			clustId = j;
			dist = d;
		}
	}
	return clustId;
}

bool Kmean::caluCenter()
{
	bool classfied = true;
	Pattern tempPattern;
	for(int i = 0; i < this->patterns.numOfCluster; i++)
	{
		tempPattern.clear();
		for(int j = 0;j < this->patterns.numOfDim; j++)
		{
			tempPattern.push_back(0);
		}
		Kmean::Cluster cluster = this->clusters[i];
		if(cluster.numOfPatterns<=0)
		{
			cluster.setCenter(tempPattern);
		}
		else
		{
			for(int j = 0; j < cluster.numOfPatterns; j++)
			{
				for(int k = 0; k < this->patterns.numOfDim; k++)
				{
					Pattern pattern = this->patterns.getPattern(cluster.getPattern(j));
					tempPattern[k] += pattern[k];
				}
			}
			for(int j = 0; j < this->patterns.numOfDim; j++)
			{
				tempPattern[j] = tempPattern[j] / cluster.numOfPatterns;
				double min = tempPattern[j] - cluster.getCenter()[j];
				if((tempPattern[j] != cluster.getCenter()[j]))
					classfied = false;
			}
			this->clusters[i].setCenter(tempPattern);
		}
	}
	return classfied;
}

void Kmean::saveClusters()
{
	char ifsave;
	cout<<"the sample has been classfied...."<<endl;
	cout<<"save the clusters?[y/n]: ";
	cin>>ifsave;
	if(ifsave=='y' || ifsave=='Y')
	{
		cout<<"input the target file name: ";
		string fileName;
		cin>>fileName;
		ofstream fout;
		fout.open(fileName.c_str(), ios::out);
		fout<<this->patterns.numOfPatterns<<endl;
		fout<<this->patterns.numOfDim<<endl;
		fout<<this->patterns.numOfCluster<<endl;
		for(int i = 0; i < this->patterns.numOfCluster; i++)
		{
			Kmean::Cluster cluster = this->clusters[i];
			/*for(int j = 0; j < cluster.numOfPatterns; j++)
			{
				Pattern pattern = this->patterns.getPattern(cluster.getPattern(j));
				for(int k = 0; k < this->patterns.numOfDim; k++)
				{
					fout<<pattern[k]<<" ";
				}
				fout<<i+1<<endl;
			}*/
			for(int j = 0; j < this->patterns.numOfDim; j++)
			{
				
				for(int k = 0; k < cluster.numOfPatterns-1; k++)
				{
					Pattern pattern = this->patterns.getPattern(cluster.getPattern(k));
					fout<<pattern[j]<<endl;
				}
				fout<<"*************************************"<<endl;
			}
		}
		fout.close();
	}
}

void Kmean::display()
{
	for(int i = 0; i < this->patterns.numOfCluster; i++)
	{
		Kmean::Cluster cluster = this->clusters[i];
		Pattern center = cluster.getCenter();
		cout<<"center: "<<center[0]<<"   "<<center[1]<<endl;
		for(int j = 0; j < cluster.numOfPatterns; j++)
		{
			
			Pattern pattern = this->patterns.getPattern(cluster.getPattern(j));
			for(int k = 0; k < this->patterns.numOfDim; k++)
			{
				cout<<pattern[k]<<" ";
			}
			cout<<i+1<<endl;
		}
	}
	cout<<endl<<endl;
}

void Kmean::run()
{
	bool classfied = false;

	this->patterns.load(KMEAN);
	this->initClusters();
	this->patterns.display();
	while(!classfied)
	{
		this->classifyPattern();
		classfied = this->caluCenter();
	}
	this->saveClusters();
}