Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- %obsluga elementow nietypowych
- %klasteryzacja
- %klasyfikacja
- %metoda bayesa iris from uci maschine learning
- clear;clc;
- load fisheriris;
- lab=findgroups(species);
- d1=3;
- d2=4;
- gscatter(meas(:,d1),meas(:,d2),lab);
- n=length(lab);
- ratio=.2;
- an=ratio*n;
- bn=n-an;
- aidx=randperm(n,an);
- bidx=setdiff(1:n,aidx);
- adata=meas(aidx,:);
- bdata=meas(bidx,:);%etykiety
- alab=lab(aidx);
- blab=lab(bidx);
- lab2=lab;
- lab2(bidx)=lab2(bidx)+3;
- gscatter(meas(:,d1),meas(:,d2),lab2);
- %korzystamy z klasyfikatora bayes w celu klasyfikacji
- for j=1:3
- mu(j,:)=mean(adata(alab==j,:));
- sigma(j,:)=std(adata(alab==j,:));
- pc(j)=sum(alab==j)/an;
- end
- for j=1:3
- for i=1:bn
- P(i,j)=pc(j)*mvnpdf(bdata(i,:),mu(j,:),sigma(j,:));
- end
- end
- [~,reslab]=max(P,[],2);
- skutecznosc=sum(blab==reslab)/bn;
- %klasyfikacja
- X=linspace(min(meas(:,d1)),max(meas(:,d1)),100);
- Y=linspace(min(meas(:,d2)),max(meas(:,d2)),100);
- [gx,gy]=meshgrid(X,Y);
- for j=1:3
- gp(:,j)=pc(j)*mvnpdf([gx(:),gy(:)],mu(j,[d1,d2]),sigma(j,[d1,d2]));
- end
- [~,glab]=max(gp,[],2);
- gscatter(gx(:),gy(:),glab);
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement