test_k_means.zip_K._k means 界面

#include "k_means.h" double get_distance(double* src, double* dest, int dimension) { double dist = 0; for(int i=0;i<dimension;i++) { dist += pow((dest[i]-src[i]),2.0); } dist = pow(dist,0.5); return dist; } int check_equal(double* src, double* dest, int cnt) { for(int i=0;i<cnt;i++) { if(abs(src[i]-dest[i])>0.000001)return 0; } return 1; } int k_means(double* data, int dimension, int count, int k, double* avg,int* cluster) { if(count<k)return -1; double* center = new double[dimension*k]; int* center_cnt = new int[k]; double* old_center = new double[dimension*k]; int* idx = new int[count]; memset(center,0,sizeof(double)*dimension*k); memset(center_cnt,0,sizeof(int)*k); memset(old_center,0,sizeof(double)*dimension*k); memset(idx,0,sizeof(int)*k); int step = count/k; for(int i=0;i<k;i++) { double* dest = center+(i*dimension); double* src = data+(i*step*dimension); memcpy(dest, src, sizeof(double)*dimension); } do { memcpy(old_center, center, sizeof(double)*dimension*k); memset(center,0,sizeof(double)*dimension*k); memset(center_cnt,0,sizeof(int)*k); for(int i=0;i<count;i++) { double min_dist = get_distance(data+(i*dimension), old_center, dimension); int min_dist_idx = 0; for(int j=1;j<k;j++) { double _dist = get_distance(data+(i*dimension),old_center+(j*dimension),dimension); if(min_dist>_dist) { min_dist = _dist; min_dist_idx = j; } } idx[i] = min_dist_idx; for(int j=0;j<dimension;j++) { center[idx[i]*dimension+j]+=data[i*dimension+j]; } center_cnt[idx[i]]+=1; } for(int i=0;i<k;i++) { for(int j=0;j<dimension;j++) { center[i*dimension+j] = center[i*dimension+j]/center_cnt[i]; } } }while(!check_equal(old_center,center,k*dimension)); if(cluster!=NULL)memcpy(cluster, idx, sizeof(int)*count); memcpy(avg, center, sizeof(double)*dimension*k); delete idx; delete old_center; delete center_cnt; delete center; return 0; }