naivedis.cpp

某个实验事编写粗糙集智能信息处理的程序

  // CNaivedisDoc.cpp: implementation of the CNaivedisDoc class.
//
//////////////////////////////////////////////////////////////////////

#include "stdafx.h"
#include "../RSet.h"
#include "fstream.h"
#include "Naivedis.h"
#include "math.h"
#include "stdlib.h"
#include "string.h"
#include "stdio.h"
//#define MAX 30


#ifdef _DEBUG
#undef THIS_FILE
static char THIS_FILE[]=__FILE__;
#define new DEBUG_NEW
#endif

//////////////////////////////////////////////////////////////////////
// Construction/Destruction
//////////////////////////////////////////////////////////////////////

CNaivedis::CNaivedis()//构造函数对变量初始化
{
	
	  Cut=NULL;
	  NewTable=NULL;
      pAttName=NULL;
      pDataType=NULL;
	  pStringTable=NULL;
      pStrResult=NULL;
      pNonStringTable=NULL;  
      pStringTableResult=NULL;
      iAttNum=0;
      iStrAttNum=NULL;
      iNonStrAttNum=NULL;
	  strCuts=NULL; 
}

CNaivedis::~CNaivedis()//析构函数释放变量空间
{
	int i;
	if(Cut!=NULL)
	{
		for(i=0;i<iNonStrAttNum;i++)
			delete []Cut[i];
		delete []Cut;
	}
	if(NewTable!=NULL)
	{
		for(i=0;i<iRecordNum;i++)
			delete []NewTable[i];
		delete []NewTable;
	}
	
	if(pAttName!=NULL)
	{
		for(i=0;i<iNonStrAttNum+1;i++)
			delete []pAttName[i];
		delete[] pAttName;
	}

	if(pDataType!=NULL)
	{
		for(i=0;i<iNonStrAttNum+1;i++)
			delete []pDataType[i];
		delete[] pDataType;
	}
	if(pStringTable!=NULL)
	{
		for(i=0;i<iRecordNum;i++)
		{
			for(int j=0;j<iStrAttNum;j++)
				delete[] pStringTable[i][j];
			delete[] pStringTable[i];
		}
		delete[] pStringTable;
	}
	if(pStringTableResult!=NULL){ 
         for(i=0;i<iRecordNum;i++) 
               delete[] pStringTableResult[i]; 
         delete[] pStringTableResult; 
	} 
    if(pStrResult!=NULL){ 
        for(i=0;i<iRecordNum;i++){ 
              for(int j=0;j<iStrAttNum;j++) 
				  if(pStrResult[i][j]!=NULL)
					  delete[] pStrResult[i][j]; 
        delete[] pStrResult[i]; 
		} 
        delete[] pStrResult; 
	} 
    if(pNonStringTable){ 
        for(i=0;i<iRecordNum;i++) 
             delete[] pNonStringTable[i]; 
        delete[] pNonStringTable; 
	} 
    if(strCuts) 
      delete[] strCuts; 
}

void CNaivedis::GetNewTable()////根据所产生的断点集得到离散化后的决策表并存入NewTable中
{
   int i,j,k;
   int m;
   for(i=0;i<iNonStrAttNum;i++)
	  GetCut(i);
   NewTable=new int *[iRecordNum];
   for(i=0;i<iRecordNum;i++)
	{
	   try
		{
         NewTable[i]=new int[iNonStrAttNum+1];
		}
		catch(CMemoryException * e)
		{
	    	::MessageBeep(MB_ICONHAND);
	       	AfxMessageBox("Out of the memory!",MB_OK|MB_ICONSTOP);
			e->Delete();
		} 
   }
   for(i=0;i<iRecordNum;i++)
   {
       for(j=0;j<iNonStrAttNum;j++)
	   {
		    m=0;
			for(k=0;k<(int)Cut[j][0];k++)
			{
				if(pNonStringTable[i][j]<Cut[j][k+1])
				{
					NewTable[i][j]=m;
					break;
				}
				m++;
			}
			   if(pNonStringTable[i][j]>=Cut[j][(int)Cut[j][0]])
					NewTable[i][j]=(int)Cut[j][0];
			}
	   NewTable[i][iNonStrAttNum]=(int)pNonStringTable[i][iNonStrAttNum];
   }

}
//具体实现离散化的地方
BOOL CNaivedis::OnNaivedis(){ 
    int i; 
    i=InitTable();    
    if(i==1)
		GetNewTable();  
    if(iStrAttNum!=0) 
        doString(); 
    return TRUE; 
} 

//保存离散化后的结果，包括离散化后的决策表和所选取的断点集
bool CNaivedis::OnNaivedisSave(LPCTSTR lpszPathName)
{
	int i,j;
	fstream fpw;
	fpw.open(lpszPathName,ios::out);
	if(!fpw)
	{
		::MessageBeep(MB_ICONHAND);
		AfxMessageBox("some error happen, file can't be opend!",
		MB_OK|MB_ICONSTOP);
        exit(0);
	}
	fpw<<"Style:"<<"train"<<endl;
	fpw<<"Stage:2"<<endl;
    fpw<<"Condition attributes number:"<<iAttNum<<endl;
    fpw<<"Records number:"<<iRecordNum<<endl;
	
	for(i = 0;i < iAttNum+1;i++)
		fpw<<pAttName[i]<<" ";
	fpw<<endl;

	for(i = 0;i < iAttNum+1;i++)
		fpw<<pDataType[i]<<" ";
	fpw<<endl;
	int strIndex=0,nonStrIndex=0; 
        for(i=0;i<iRecordNum;i++){ 
        for(j=0;j< iAttNum;j++){ 
            if(!strcmp(pDataType[j],"String")){ 
                fpw<<pStringTableResult[i][strIndex++]<<"   "; 
            } 
            else{ 
                fpw<<NewTable[i][nonStrIndex++]<<"  "; 
            } 
        } 
        fpw<<NewTable[i][nonStrIndex];//决策属性 
        strIndex=0,nonStrIndex=0; 
        fpw<<endl; 
    } 
	
    fpw<<"[Cuts]"<<endl;
    strIndex=0,nonStrIndex=0; 
    for(i=0;i<iAttNum;i++){ 
        fpw<<i<<endl; 
        if(!strcmp(pDataType[i],"String")){ 
            fpw<<strCuts[strIndex]<<endl;   //断点个数 
            for(j=0;j<strCuts[strIndex];j++){ 
                fpw<<pStrResult[j][strIndex]<<" "<<j<<endl; //断点与离散值对应关系 
            } 
            strIndex++; 
        } 
        else{ 
            fpw<<Cut[nonStrIndex][0]+1<<endl;   //断点个数 
            fpw<<"["<<"*"<<","; 
            for(j=0;j<(int)Cut[nonStrIndex][0];j++){ 
                fpw<<Cut[nonStrIndex][j+1]<<")"<<"  "<<j<<endl; 
                fpw<<"["<<Cut[nonStrIndex][j+1]<<","; 
            } 
            fpw<<"*"<<"]"<<" "<<j; 
            fpw<<endl; 
            nonStrIndex++; 
        } 
    } 
    fpw.close(); 
	return true;
}
//给cut数组(保存断点的数组)分配内存
int CNaivedis::InitTable()
{
     int i=0,j=0;

	Cut=new float*[iAttNum];
	for(i=0;i<iAttNum;i++)
	{
	   try
		{
         Cut[i]=new float[iRecordNum];
		}
		catch(CMemoryException * e)
		{
	    	::MessageBeep(MB_ICONHAND);
	       	AfxMessageBox("Out of the memory!",MB_OK|MB_ICONSTOP);
			e->Delete();
	    	return -2;
		} 
	}
	
	return  1;

}
//先获得第i个属性的属性值，再对它进行排序，最后得到这个属性上的断点保存在数组cut[i][]中
void CNaivedis::GetCut(int i)
{
   int j,k=1;
   float **Mid;
   Cut[i][0]=0;
   Mid=new float*[iRecordNum];
   for(j=0;j<iRecordNum;j++)
	{
	   try
		{
         Mid[j]=new float[2];
		}
		catch(CMemoryException * e)
		{
	    	::MessageBeep(MB_ICONHAND);
	       	AfxMessageBox("Out of the memory!",MB_OK|MB_ICONSTOP);
			e->Delete();
		} 
   }	   
   for(j=0;j<iRecordNum;j++)
   {
     Mid[j][0]=pNonStringTable[j][i];//属性值
	 Mid[j][1]=pNonStringTable[j][iNonStrAttNum];//决策值
   }
   SelectSort(Mid);
   for(j=0;j<iRecordNum-1;j++)
	  if((Mid[j][0]!=Mid[j+1][0])&&(Mid[j][1]!=Mid[j+1][1]))
	  {	  
	      Cut[i][k++]=(Mid[j][0]+Mid[j+1][0])/2;
		  Cut[i][0]++;
	  }
   for(j=0;j<iRecordNum;j++)
	   delete []Mid[j];
   delete []Mid;
}
//对某个属性进行排序，输入没有排序的数组Mid，输出排序后数组Mid
void CNaivedis::SelectSort(float **Mid)
{
   int i,j;
   float Mid1;
   float Mid2;
   for(i=0;i<iRecordNum-1;i++)
	   for(j=i+1;j<iRecordNum;j++)
		   if(Mid[i][0]>Mid[j][0])
		   {
			   Mid1=Mid[i][0];
			   Mid2=Mid[i][1];
			   Mid[i][0]=Mid[j][0];
		       Mid[i][1]=Mid[j][1];  
			   Mid[j][0]=Mid1;
			   Mid[j][1]=Mid2;
		   }
}
//从文件中读取决策表的所有信息
BOOL CNaivedis::ReadDataFromFile(char *filename){ 
    FILE *fp; 
    if((fp = fopen(filename,"r")) == NULL){ 
        ::MessageBeep(MB_ICONHAND); 
        AfxMessageBox("Couldn't open the file",MB_OK|MB_ICONSTOP); 
        return FALSE; 
    }//end if 
    fscanf(fp,"Style:%s\n",cStyle); 
    fscanf(fp,"Stage:%d\n",&iStage); 
    fscanf(fp,"Condition attributes number:%d\n",&iAttNum); 
    if(_stricmp(cStyle,"train") == 0 && iStage<=1){ 
        fscanf(fp,"Records number:%d\n",&iRecordNum); 
        if(!readAttrInfo(fp)) 
            return FALSE; 
        if(!readTable(fp)) 
            return FALSE; 
    } 
    return TRUE; 
} 
//读取属性名到pAttName中,将数据类型读入pDataType中
BOOL CNaivedis::readAttrInfo(FILE* fp){ 
    int i; 
    if(pAttName == NULL){ 
        try{ 
            pAttName = new char*[iAttNum+1]; 
            pDataType= new char*[iAttNum+1]; 
        } 
        catch(CMemoryException* e){ 
            ::MessageBeep(MB_ICONHAND); 
            AfxMessageBox("Out of memory!",MB_OK|MB_ICONSTOP); 
            e->Delete(); 
            return FALSE; 
        } 
        for(i=0;i <= iAttNum;i++){ 
            try{ 
                pAttName[i]=new char[MAX]; 
                pDataType[i]=new char[MAX]; 
            } 
            catch(CMemoryException* e){ 
                ::MessageBeep(MB_ICONHAND); 
                AfxMessageBox("Out of memory!",MB_OK|MB_ICONSTOP); 
                e->Delete(); 
                return FALSE; 
            } 
        }//end for 
    }//end if 
    for(i=0;i <= iAttNum;i++) 
        fscanf(fp,"%s",pAttName[i]); 
    fscanf(fp,"\n"); 
        //读属性名 
    iStrAttNum = 0; 
    iNonStrAttNum = 0; 
    for(i = 0;i < iAttNum;i++){ 
        fscanf(fp,"%s",pDataType[i]); 
        if(!strcmp(pDataType[i],"String")) 
            iStrAttNum++;//字符串属性的个数 
        else 
            iNonStrAttNum++; //非字符串属性个数
    }//读属性值类型 
    fscanf(fp,"%s",pDataType[iAttNum]); 
        //决策属性类型 
    fscanf(fp,"\n"); 
    return TRUE; 
} 
//读入属性值
BOOL CNaivedis::readTable(FILE* fp){ 
    int i,j; 
    if((pStringTable == NULL)&&(pNonStringTable == NULL)){ 
        try 
        { 
            if(iStrAttNum!=0) 
                pStringTable = new char**[iRecordNum]; 
            pNonStringTable = new float*[iRecordNum]; 
                //至少决策属性是保存在pNonStringTable中的,所以pNonStringTable一 
//定不为空 
        } 
        catch(CMemoryException* e){ 
            ::MessageBeep(MB_ICONHAND); 
            AfxMessageBox("Out of memory!",MB_OK|MB_ICONSTOP); 
            e->Delete(); 
            return FALSE; 
        } 
        for(i = 0;i < iRecordNum;i++){ 
            try{ 
                if(iStrAttNum!=0) 
                    pStringTable[i] = new char*[iStrAttNum]; 
                pNonStringTable[i] = new float[iNonStrAttNum+1];//包括决策属性 
            } 
            catch(CMemoryException* e){ 
                ::MessageBeep(MB_ICONHAND); 
                AfxMessageBox("Out of memory!",MB_OK|MB_ICONSTOP); 
                e->Delete(); 
                return FALSE; 
            } 
            for(j = 0;j < iStrAttNum;j++){ 
                try{ 
                    pStringTable[i][j] = new char[MAX]; 
                } 
                catch(CMemoryException* e){ 
                    ::MessageBeep(MB_ICONHAND); 
                    AfxMessageBox("Out of memory!",MB_OK|MB_ICONSTOP); 
                    e->Delete(); 
                    return FALSE; 
                } 
            }//end for 
            for(j=0;j<=iNonStrAttNum;j++) 
                pNonStringTable[i][j]=0.0; 
        }//end for 
    }//end if 
    int iStrIndex,iNonStrIndex; 
	iStrIndex=0;
	iNonStrIndex=0;
    char* str; 
    str=new char[MAX]; 
    for(i = 0;i < iRecordNum;i++){ 
        for(j = 0;j < iAttNum;j++){ 
            if(!strcmp(pDataType[j],"String")) 
              	fscanf(fp,"%s",pStringTable[i][iStrIndex++]); 
		     //字符串类型的属性值读入pStringTable 
            else{ 
                fscanf(fp,"%s",str); 
                pNonStringTable[i][iNonStrIndex++]=(float)atof(str); 
            }//非字符串类型的属性值读入pNonStringTable 
        } 
        fscanf(fp,"%s",str);//读入决策属性的值 
        fscanf(fp,"\n"); 
        pNonStringTable[i][iNonStrIndex]=(int)atof(str); 
        iStrIndex=0,iNonStrIndex=0; 
    } 
    delete[] str; 
    return TRUE; 
} 
void CNaivedis::doString(){ 
    //先建立一个pStrResult,通过对pStringTable的一次扫描,把每一个属性的所有可能值 
//找出来,建立字符串值与离散值的映射关系 
    //再把pStringTable中的数据与pStrResult中的数据进行比较,修改pStringTable中的 
//值,以便打印结果 
    //strCuts[]存放断点个数 
    int i=0,j=0,k=0; 
    try{ 
        pStrResult = new char**[iRecordNum]; 
        pStringTableResult = new int*[iRecordNum]; 
            //存放离散化以后的结果 
        strCuts = new int[iStrAttNum]; 
            //存放字符串属性的断点个数 
    } 
    catch(CMemoryException* e){ 
        ::MessageBeep(MB_ICONHAND); 
        AfxMessageBox("Out of memory!",MB_OK|MB_ICONSTOP); 
        e->Delete(); 
    } 
    for (i=0;i<iRecordNum;i++){ 
        try{ 
            pStrResult[i]= new char*[iStrAttNum]; 
            pStringTableResult[i]= new int[iStrAttNum]; 
        } 
		catch(CMemoryException* e){ 
            ::MessageBeep(MB_ICONHAND); 
            AfxMessageBox("Out of memory!",MB_OK|MB_ICONSTOP); 
            e->Delete(); 
        } 
		try{
         for(j=0;j<iStrAttNum;j++){ 
            pStrResult[i][j] = new char[20]; 
		//    pStrResult[i][j] =NULL;//不用直接赋值方式,因为这样后面就不能拷贝再赋值
			strcpy(pStrResult[i][j],"");//用拷贝方式给字符串赋值
        }
		}
		catch(CMemoryException* e){ 
            ::MessageBeep(MB_ICONHAND); 
            AfxMessageBox("Out of memory!",MB_OK|MB_ICONSTOP); 
            e->Delete(); 
        } 
    } 
    for(i=0;i<iStrAttNum;i++){ 
        for(j=0;j<iRecordNum;j++) 
            pStringTableResult[j][i]=0; 
        strCuts[i]=0; 
    } 
    for(i=0;i<iStrAttNum;i++){ //i列
        for(j=0;j<iRecordNum;j++){//对每个属性值 
            for(k=0;k<=j; k++){ 
                if(!strcmp(pStringTable[j][i],pStrResult[k][i])) 
                    break; 
                else if(!strcmp(pStrResult[k][i],"")){ 
//如果pStrResult[k][i]不等于pStringTable[j][i]而且pStrResult[k][i]为空 
//则应该把pStringTable[j][i]赋值给pStrResult[k][i]
                    strcpy(pStrResult[k][i],pStringTable[j][i]); //j样例i属性
	                strCuts[i]++;//记录断点个数 
                    break; //退出k循环
				} //end else if
            } //end for(k)
            pStringTableResult[j][i]=k;
//把pStringTableResult[j][i]的值改为离散化以后的值 
        } //end for(j)
    } //end for(i)
	
}