决策树算法C4.5 

数据仓库与数据挖掘

iGet = getc(fp); } i++;

iAttribute = (int *)malloc(sizeof(int)*i); for (int k = 0; k

iAttribute[k] = (int)malloc(sizeof(int)); iAttribute[k] = 1; }

while (EOF != iGet)

{

if ('\\n' == iGet) {

j++; }

iGet = getc(fp); } j++;

iInput = (int **)malloc(sizeof(int*)*j); iSamples = (int *)malloc(sizeof(int)*j); for (a = 0;a < j;a++) {

iInput[a] = (int *)malloc(sizeof(int)*i); iSamples[a] = (int)malloc(sizeof(int)); iSamples[a] = a; }

a = 0;

fclose(fp);

fp=fopen(\ iGet = getc(fp); while(EOF != iGet) {

if ((',' != iGet)&&('\\n' != iGet)) {

iInput[a][b] = iGet - 48; b++; }

if (b == i) {

a++;

数据仓库与数据挖掘

b = 0; }

iGet = getc(fp); }

fp1 = fopen(\

build_tree(fp1,iSamples,iAttribute,0); fclose(fp); return 0; }

void build_tree(FILE * fp, int* iSamples, int* iAttribute,int level)// {

int iTest_Attribute = 0; int iAttributeValue[MAX]; int k = 0; int l = 0; int m = 0;

int *iSamples1;

for (k = 0; k

{

iAttributeValue[k] = -1; }

if (0 == check_samples(iSamples)) {

fprintf(fp,\ return; }

if (1 == check_attribute_null(iAttribute)) {

fprintf(fp,\ return; }

iTest_Attribute = choose_attribute(iSamples,iAttribute); iAttribute[iTest_Attribute] = -1;

get_attributes(iSamples,iAttributeValue,iTest_Attribute);

数据仓库与数据挖掘

k = 0;

while ((-1 != iAttributeValue[k])&&(k < MAX)) {

l = 0; m = 0;

while ((-1 != iSamples[l])&&(l < j)) {

if (iInput[iSamples[l]][iTest_Attribute] == iAttributeValue[k]) {

m++; } l++; }

iSamples1 = (int *)malloc(sizeof(int)*(m+1)); l = 0; m = 0;

while ((-1 != iSamples[l])&&(l < j)) {

if (iInput[iSamples[l]][iTest_Attribute] == iAttributeValue[k]) {

iSamples1[m] = iSamples[l]; m++; } l++; }

iSamples1[m] = -1;

if (-1 == iSamples1[0]) {

fprintf(fp,\ return; }

fprintf(fp,\

= %d\\n\ build_tree(fp,iSamples1,iAttribute,level+1); k++; } }

int choose_attribute(int* iSamples, int* iAttribute) {

int iTestAttribute = -1; int k = 0; int l = 0; int m = 0;

数据仓库与数据挖掘

int n = 0;

int iTrue = 0; int iFalse = 0; int iTrue1 = 0; int iFalse1 = 0; int iDepart[MAX]; int iRecord[MAX];

double dEntropy = 0.0; double dGainratio = 0.0; double test = 0.0;

for (k = 0;k

iDepart[k] = -1; iRecord[k] = 0; }

k = 0;

while ((l!=2)&&(k<(i - 1))) {

if (iAttribute[k] == -1) {

l++; } k++; }

if (l == 1)

{

for (k = 0;k<(k-1);k++) {

if (iAttribute[k] == -1) {

return iAttribute[k]; } } }

for (k = 0;k < (i-1);k++) {

l = 0;

iTrue = 0; iFalse = 0;

if (iAttribute[k] != -1) {

数据仓库与数据挖掘

while ((-1 != iSamples[l])&&(l < j)) {

if (0 == iInput[iSamples[l]][i-1]) {

iFalse++; }

if (1 == iInput[iSamples[l]][i-1]) {

iTrue++; }

l++; }

for (n = 0;n

m = 0;

while((iDepart[m]!=-1)&&(m!=MAX)) {

if (iInput[iSamples[n]][iAttribute[k]] == iDepart[m]) {

break; } m++; }

if (-1 == iDepart[m]) {

iDepart[m] = iInput[iSamples[n]][iAttribute[k]]; } }

while ((iDepart[m] != -1)&&(m!=MAX)) {

for (n = 0;n

if (iInput[iSamples[n]][iAttribute[k]] == iDepart[m]) {

if (1 == iInput[iSamples[n]][i-1]) {

iTrue1++; }

if (0 == iInput[iSamples[n]][i-1]) {

iFalse1++; }

iRecord[m]++; }