MATLAB: Rectify error in kmeans

errormammogram

i wanted to do kmeans clustering in mammogram image…. i took the datavector of the image and gave as input to kmeans…. when i execute the code error is displayed as…."An empty cluster error occurred in every replicate."…. what should i do to rectify this error….

Best Answer

The kmeans function has an option 'EmptyAction' that you could try. With this option you can request that, if a cluster becomes empty, the function either accept a smaller number of clusters or create a new one from a point that is relatively far from the others.

Related Solutions

MATLAB: Kmeans

You do the same thing except:

    plot(X(idx==1,1),X(idx==1,2),'r.','MarkerSize',10);
    hold on
    plot(X(idx==2,1),X(idx==2,2),'b.','MarkerSize',10);
    plot(X(idx==3,1),X(idx==3,2),'g.','MarkerSize',10);
    plot(X(idx==4,1),X(idx==4,2),'k.','MarkerSize',10);

MATLAB: Clustering by kmean

See the documentation on the kmeans function.

IDX = kmeans(X,K)

The first input (X) should be an N-by-P data matrix, where N is the number of instances to cluster and P is the number of features. The second input (K) is the number of clusters you desire. The output (IDX) will be an N-by-1 vector of cluster indices for each data point.

In your script, you call

idx = kmeans(8759,52);

The first input is the problem. It is a scalar! In other words, you have only 1 data point with 1 feature. It is impossible to cluster this "data" into 52 clusters. From a very brief glance at your code, I think you may want this instead:

idx = kmeans(f,52);

Best Answer

Related Solutions

MATLAB: Kmeans

MATLAB: Clustering by kmean

Related Question