clustering.cpp

将Weka数据挖掘工具所产生的K-MEANS和DBSCAN结果转化成MATLAB可输出三维图像的格式

#include<stdio.h>  
#include <iostream>
  #include<string>   
  #include<fstream>   
#include <sstream> 

  using   namespace   std;   
 
 void DBSCAN(); 
 void K_means();
 void main()   
  {   
	 int algorithm=1;
	 cout<<"Choose clustering algorithm:"<<endl;
	 cout<<"DBSCAN:	1"<<endl;
	 cout<<"K-means:2"<<endl;
	 cout<<"your choice is:";
	 cin>>algorithm;
	 cout<<algorithm<<endl;
	 if(algorithm==1)
		 DBSCAN();
	 else if(algorithm==2)
		 K_means();

		  
		  
  }

 void K_means()
 {
	ifstream infile;
	string str;
	//float weight=1.2;

	//对象数目
	int num_of_Objects;

	//聚类数目
	int num_of_Clusters;

	//聚类中心
	//float *x_cent;
	//float *y_cent;
	//float *z_cent;
	
	float * x;
	float * y;
	float * z;
	int * cluster_num;
	
	//属性数量
	 int num_of_Attributes=0;
	int flag_record_Attributes_num=0;

	//实际使用属性数量
	 int use_num_Attributes=0;

	 //记录每个聚类中的对象数目
	 //int * objects_in_Cluster;

	cout<<"begin Kmeans"<<endl;
	infile.open("a.txt"); 
	 while(!infile.eof())
	 {
		 infile>>str;
		/* if(str.compare("Attributes:")==0&&flag_record_Attributes_num!=1)
		 {
			 infile>>num_of_Attributes;
			 flag_record_Attributes_num=1;
			 int temp_Attributes=0;
			 string *Attributes=new string[num_of_Attributes];
			 while(temp_Attributes<num_of_Attributes)
			 {
				 
				 infile>>str;
				 if(str.compare("Ignored:")==0)
					use_num_Attributes=temp_Attributes;
				 else
				 {
					Attributes[temp_Attributes]=str;
					//cout<<Attributes[temp_Attributes]<<endl;
					temp_Attributes++;
				 }
				 
			 }
			 
		 }*/
		 //读出对象数目
		 //else 
		if(str.compare("Instances:")==0)
		{
			 infile>>num_of_Objects;
			 cout<<"num objects:"<<num_of_Objects<<endl;
			 
		 }
		 //读出聚类数目
		 else if(str.compare("-N")==0)
		 {
			 infile>>num_of_Clusters;
		/*	 x_cent=new float[num_of_Clusters];
			 y_cent=new float[num_of_Clusters];
			 z_cent=new float[num_of_Clusters];
			 objects_in_Cluster=new int[num_of_Clusters];*/
			 cout<<"num_of_Clusters:"<<num_of_Clusters<<endl;
		 }
		/* else if(str.compare("Cluster")==0)
		 {
			 infile>>str;
			 if(str.compare("centroids:")==0)
			 {
				 for(int i=0;i<num_of_Clusters;i++)
				 {
					 infile>>str;
					 infile>>str;
					 infile>>str;
					 infile>>x_cent[i];
					 //x_cent[i]=x_cent[i]
					 infile>>y_cent[i];
					 infile>>z_cent[i];
					 cout<<x_cent[i]<<" "<<y_cent[i]<<" "<<z_cent[i]<<endl;
					 
				 }
			 }


		 }*/
	 }
	 infile.close();
	 ifstream infile_1;

	 x=new float[num_of_Objects];
	 y=new float[num_of_Objects];
	 z=new float[num_of_Objects];
	 x[0]=1;
	 cluster_num=new int[num_of_Objects];
	 infile_1.open("a.arff");
	 int index=0;
	 //读出坐标和所属类的编号
	 
	 while(!infile_1.eof())
	 {
		 infile_1>>str;
		 
		 if(str.compare("@data")==0)
		 {
			 for(int i=0;i<num_of_Objects;i++)
			 {
				 infile_1>>index;
				 cout<<"index:"<<index<<endl;
				 infile_1>>x[i];
				 cout<<"x[0]:"<<x[i]<<endl;
				 infile_1>>y[i];
				 infile_1>>z[i];

				 infile_1>>str;
				 infile_1>>str;
				 infile_1>>str;
				 infile_1>>str;
				 infile_1>>str;
				 infile_1>>str;
				 infile_1>>str;
				 infile_1>>str;
				 infile_1>>str;
				 infile_1>>str;
				 infile_1>>str;
				 infile_1>>str;
				 infile_1>>str;
				 infile_1>>str;
				 
				 char temp=str.at(7);
				 cluster_num[index]=atoi(&temp);
				 //cout<<clusterNum[index]<<endl;
			 }
		 }
	 }
	 infile_1.close();
	 ofstream outfile; 
	 outfile.open("result_kmeans.m");
	  //输出“hold on”		
	 str="hold on\n";
	 outfile<<str;

	 //画点的样式
	 string pattern[100]={"k.","r^","o","b+","r*","ks","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<"};

	  

	 
	
	 //输出各个点的坐标值
	 /*例如：	x = [52 43 45];
				y = [1 4 4];
				z= [0 2 2];
				x1=[62 51 54 46]
				y1=[3 0 0 0]
				z1=[3 0 0 0]
		*/
	 for(int i=0;i<num_of_Objects;i++)	
	 {
		 outfile<<"plot3("<<x[i]<<","<<y[i]<<","<<z[i]<<",'"<<pattern[cluster_num[i]]<<"')"<<endl;
	 }

	

	



	 str="hold off\nbox on\ngrid on\n";
	 outfile<<str;
	 
	 //输出坐标轴定义
	 str="axis([30 90 0 6 0 6])\nxlabel('age')\nylabel('BIRADS-L')\n";
	 outfile<<str;
	 str="zlabel('Focus_Num')";
	 outfile<<str;
	 outfile.close();

	  


 }

 void DBSCAN()
 {
	 ifstream infile; 
	 


	 //属性数量
	 int num_of_Attributes=0;
	int flag_record_Attributes_num=0;

	 //实际使用属性数量
	 int use_num_Attributes=0;

	 //对象数目
	 int num_of_Objects=0;


	 //DBSCAN算法参数
	 float Epsilon;
	 int Minpts;

	 //Cluster数目
	 int num_of_Clusters=0;

	 //噪声数目
	 int num_of_Noise=0;

	 //算法耗时
	 float time;

	 //坐标值
	 float * x;
	 float * y;
	 float * z;
	 //记录每个对象所属的聚类编号
	 int * cluster_num;
	 //记录每个聚类中的对象数目
	 int * objects_in_Cluster;

	 int flag_record_xyz=0;
		cout<<"begin dbscan"<<endl;
	 infile.open("boca-focumNum.txt"); 
	 
	 string str;
	 string *temp_Objects;

	 float weightx=1;
	 float weighty=10;
	 float weightz=10;
	 
//	 int i=4000;
	 
	 
	while(!infile.eof())
	 {
	//	i--;
		 infile>>str;
	//	 cout<<str<<endl;
			
		 //读出属性名和属性数量
		 if(str.compare("Attributes:")==0&&flag_record_Attributes_num!=1)
		 {
			 infile>>num_of_Attributes;
			 flag_record_Attributes_num=1;
			 int temp_Attributes=0;
			 string *Attributes=new string[num_of_Attributes];
			 while(temp_Attributes<num_of_Attributes)
			 {
				 
				 infile>>str;
				 if(str.compare("Ignored:")==0)
					use_num_Attributes=temp_Attributes;
				 else
				 {
					Attributes[temp_Attributes]=str;
					cout<<Attributes[temp_Attributes]<<endl;
					temp_Attributes++;
				 }
				 
			 }
			 
		 }
		 //得到对象数目
		 else if(str.compare("Clustered")==0)
		 {
			 infile>>str;
			 if(str.compare("DataObjects:")==0)
			 {
				infile>>num_of_Objects;
				temp_Objects=new string[num_of_Objects];

				x=new float[num_of_Objects];
				y=new float[num_of_Objects];
				z=new float[num_of_Objects];
				cluster_num=new int[num_of_Objects];
				int j;
				char temp[10];
				 for(j=0;j<num_of_Objects;j++)
				 {
					 itoa(j,temp,10);
					 temp_Objects[j].append(temp);
				 }

			 }
			 else if(str.compare("Instances")==0)
			{
			
				int index;
				//读出每个聚类中的对象数目
				for(int i=0;i<num_of_Clusters;i++)
				{
					infile>>index;
					infile>>objects_in_Cluster[index];
					cout<<"objects in clusters:"<<index<<" "<<objects_in_Cluster[index]<<endl;
					infile>>str;
					infile>>str;
				}
				
			}

		 }
		 //得到Epsilon值
		 else if(str.compare("Epsilon:")==0)
			 infile>>Epsilon;
		 else if(str.compare("minPoints:")==0)
			 infile>>Minpts;
		 else if(str.compare("Number")==0)
		 {
			 infile>>str;
			 if(str.compare("of")==0)
			 {
				 infile>>str;
				 if(str.compare("generated")==0)
				 {
					 infile>>str;
						if(str.compare("clusters:")==0)
						{
							infile>>num_of_Clusters;
							objects_in_Cluster=new int[num_of_Clusters];
							
						}
				 }

			 }
		 }

		 else if(str.compare("time:")==0)
			 infile>>time;
		 //编号为10之前的对象解析
		 else if(num_of_Objects>0)
		 {
			 int j=0;
			 
			 for(j=10;j<num_of_Objects;j++)
			 {
			//	 cout<<"str:"<<"("+temp_Objects[j]+".)"<<endl;
				 if(j>=10&&str.compare("("+temp_Objects[j]+".)")==0)
				 {

					 infile>>str;

					 int index=-1;
				
					 while(true)
					 {
						 index=str.find_first_of(",",index+1);
						
						 if(index==str.npos)
							 break;
						 else
							 str.replace(index,1," ");
					 }
					  
					 
					 
					 istringstream is(str);
					 is>>x[j]>>y[j]>>z[j];
					 cout<<x[j]<<" "<<y[j]<<" "<<z[j]<<endl;
					 x[j]=x[j]/weightx;
					 y[j]=y[j]/weighty;
					 z[j]=z[j]/weightz;
					 infile>>str;
					 
					 if(str.compare("-->")==0)
					 {
						 infile>>str;
						 //噪声
						 if(str.compare("NOISE")==0)
						 {
							 cluster_num[j]=-1;
							 cout<<"noise"<<endl;
						 }
						
						 //聚类编号
						 else
						 {
							 cluster_num[j]=atoi(str.c_str());
							 cout<<"num_cluster:"<<cluster_num[j]<<endl;

						 }
					 }
					 // flag_record_xyz=1;
				 }
			 }
			
		 }
		 
		 //编号为10之前的对象解析
		 if(num_of_Objects>0&&str.compare("(")==0)
		 {
			 infile>>str;
			 
			 int temp;
			 if(str.compare(1,1,".")==0)
			 {
				 
				 temp= atoi(str.replace(1,1,"").c_str());
				 cout<<temp<<endl;
				 infile>>str;

				 int index=-1;
				 while(true)
				 {
					 index=str.find_first_of(",",index+1);
					
					 if(index==str.npos)
						 break;
					 else
						 str.replace(index,1," ");
				 }
				  
				 
				 
				 istringstream is(str);
				 is>>x[temp]>>y[temp]>>z[temp];
				 x[temp]=x[temp]/weightx;
				 y[temp]=y[temp]/weighty;
				 z[temp]=z[temp]/weightz;
				 cout<<x[temp]<<" "<<y[temp]<<" "<<z[temp]<<endl;
				 infile>>str;
				 
				 if(str.compare("-->")==0)
				 {
					 infile>>str;
					 //噪声
					 if(str.compare("NOISE")==0)
					 {
						 cluster_num[temp]=-1;
						 cout<<"noise"<<endl;
					 }
					
					 //聚类编号
					 else
					 {
						 cluster_num[temp]=atoi(str.c_str());
						 cout<<"num_cluster:"<<cluster_num[temp]<<endl;
					 }
				 }
			 }
		 }
		 if(str.compare("Unclustered")==0)
		 {
			 infile>>str;
			 if(str.compare("instances")==0)
			 {
				 infile>>str;
				 infile>>num_of_Noise;
				 cout<<"num_of_noise:"<<num_of_Noise<<endl;
			 }
		 }
	 }
		cout<<"num_of_Attributes:"<<num_of_Attributes<<endl;
		cout<<use_num_Attributes<<endl;
		cout<<"num_of_Objects:"<<num_of_Objects<<endl;
		cout<<"Epsilon:"<<Epsilon<<endl;
		cout<<"Minpts:"<<Minpts<<endl;
		cout<<"num_of_Clusters:"<<num_of_Clusters<<endl;
		cout<<"Time cost:"<<time<<endl;
	
	 infile.close();
	 ofstream outfile; 
	 outfile.open("result.m");

	 float *draw_pixel_x=new float[num_of_Objects];
	 float *draw_pixel_y=new float[num_of_Objects];
	 float *draw_pixel_z=new float[num_of_Objects];
	 
	 
	 int i,j;
	 char temp[10];
	 //输出各个点的坐标值
	 /*例如：	x = [52 43 45];
				y = [1 4 4];
				z= [0 2 2];
				x1=[62 51 54 46]
				y1=[3 0 0 0]
				z1=[3 0 0 0]
		*/
	 for(i=0;i<num_of_Clusters;i++)
	 {
		 int index=0;
		 for(j=0;j<num_of_Objects;j++)
		 {
			 if(cluster_num[j]==i)
			 {
				 draw_pixel_x[index]=x[j];
				 draw_pixel_y[index]=y[j];
				 draw_pixel_z[index]=z[j];
				 cout<<draw_pixel_x[index]<<endl;
				 index++;				 
			 }
		 }
		 

		 //输出X坐标的值
		 str="x";
		 itoa(i,temp,10);

		 //使str的值为x0,x1...
		 str.append(temp);
		 str.append(" =[");
		 outfile<<str<<" ";

		 for(j=0;j<objects_in_Cluster[i];j++)
		 {
			 //float转换成string
			 sprintf(temp,"%f",draw_pixel_x[j]);
			 str=temp;

			 //X坐标上的最后一个元素
			 if(j==objects_in_Cluster[i]-1)
				outfile<<str<<"];\n";
			 //其它元素
			 else
				outfile<<str<<" ";
		 }

		 //输出Y坐标的值
		 str="y";
		 itoa(i,temp,10);

		 //使str的值为y0,y1...
		 str.append(temp);
		 str.append(" =[");
		 outfile<<str<<" ";

		 for(j=0;j<objects_in_Cluster[i];j++)
		 {
			 //float转换成string
			 sprintf(temp,"%f",draw_pixel_y[j]);
			 str=temp;

			 //Y坐标上的最后一个元素
			 if(j==objects_in_Cluster[i]-1)
				outfile<<str<<"];\n";
			 //其它元素
			 else
				outfile<<str<<" ";
		 }

		 //输出Z坐标的值
		 str="z";
		 itoa(i,temp,10);

		 //使str的值为z0,z1...
		 str.append(temp);
		 str.append(" =[");
		 outfile<<str<<" ";

		 for(j=0;j<objects_in_Cluster[i];j++)
		 {
			 //float转换成string
			 sprintf(temp,"%f",draw_pixel_z[j]);
			 str=temp;

			 //Z坐标上的最后一个元素
			 if(j==objects_in_Cluster[i]-1)
				outfile<<str<<"];\n";
			 //其它元素
			 else
				outfile<<str<<" ";
		 }
	 }

	 //输出“hold on”		
	 str="hold on\n";
	 outfile<<str;

	 //画点的样式
	 string pattern[100]={"k.","r^","o","b+","r*","ks","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<",
	 "k.","r^","go","b+","*","s","x","+",">","<"};

	 //画三维点
	 /*例如：
			plot3(x0,y0,z0,'.')
			plot3(x1,y1,z1,'^')
			plot3(x2,y2,z2,'*')
			plot3(x3,y3,z3,'.')
			plot3(x4,y4,z4,'+')
			plot3(x5,y5,z5,'*')
			*/
	 for(i=0;i<num_of_Clusters;i++)
	 {		
		 str="plot3(";
		 outfile<<str;		 
		 str="x"; 
		 itoa(i,temp,10);		
		 str.append(temp);
		 str.append(",y");
		 str.append(temp);
		 str.append(",z");
		 str.append(temp);
		 str.append(",'");
		 str.append(pattern[i]);
		 str.append("')\n");		 
		 outfile<<str;		 
	 }

	 str="hold off\nbox on\ngrid on\n";
	 outfile<<str;
	 
	 //输出坐标轴定义
	 str="axis([30 90 0 6 0 6])\nxlabel('age')\nylabel('BIRADS-L')\n";
	 outfile<<str;
	 str="zlabel('Focus_Num')";
	 outfile<<str;

	
 }