Dbscan algorithm C + + implementation

Tag:dbscan    Data mining     density     based;

#include <bits/stdc++.h> #define Dimense 10//10 Dimensional data # define N 5005#define MAX 0xffffff#define CLR (a) memset (a,0, sizeof (a)) using namespace std;double radius=60;int min_num=400;int num=5000;//Data volume int K;int now=0;//is currently the first cluster int ans=0;//    Number of clusters that are not outliers, struct point{int id;    Double Dir[dimense];    int belong;    BOOL IsKey;    BOOL Vis; int rec;//Statistics How many points are in their region vector<int>area;};    Point center[15];vector<point>data;vector<int>save[15];d ouble dis (point a,point b) {double ret=0;    for (int ii=0;ii<dimense;ii++) {ret+= (A.dir[ii]-b.dir[ii]) * (A.dir[ii]-b.dir[ii]); } return sqrt (ret);}    int cnt=0;//The total number of visited void DBscan (int cur) {int i,j;    if (Data[cur].vis) return;    Data[cur].vis=true;    Data[cur].belong=now;    cnt++;        for (i=0;i<num;i++) {if (Data[i].vis) continue;            if (DIS (data[i],data[cur]) <radius) {data[cur].area.push_back (i);            Data[i].belong=now;        data[cur].rec++; }} if (Data[cur].rec>min_num) {data[cur].iskey=true;        for (i=0;i<data[cur].rec;i++) {DBscan (data[cur].area[i]); }}else Data[cur].iskey=false;}    Set<int>tot;int vis2[n];void dfs (int a) {if (Vis2[a]) return;    Vis2[a]=1;    Save[now].push_back (a); for (int i=0;i<data[a].rec;i++) DFS (Data[a].area[i]);}    int main () {CLR (VIS2);    int i,j;    Freopen ("In.txt", "R", stdin);    Freopen ("Out2.txt", "w", stdout);        for (i=0;i<num;i++) {point T;        T.vis=false;        T.id=i;        t.rec=0;        for (j=0;j<dimense;j++) scanf ("%lf", &t.dir[j]);    Data.push_back (t);            } for (i=0;i<num;i++) if (!data[i].vis) {DBscan (i);        now++;    } now=0;            for (i=0;i<num;i++) {if (!vis2[i]) {DFS (i);        now++;    }} ans=0;    int tmp_rec[n];    for (i=0;i<now;i++) {if (Save[i].size () >min_num) tmp_rec[ans++]=i;    } printf ("Altogether%d clusters \ n Total:%d points \ n", ans,cnt); for (i=0;i<ans;i++) {int len=save[tmp_rec[i]].size ();        printf ("%d points in%d clusters", i+1,len); for (j=0;j<len;j++) {printf ("%d%c", save[tmp_rec[i]][j],j==len-1? ')        \ n ': '); }} return 0;}

Copy to Google TranslateTranslation Results

Dbscan algorithm C + + implementation