Hi, This is a problem of K-Means Clustering. I believe you have successfully extracted the features and want to start on the actual K-Means clustering. Although giving a full description is very difficult in this small post but you can do something like this :-
Suppose you have 100 data points to classify. Choose a value of K (let us say 5). Now cyclically move through all the data points and assign them to clusters (Like first data point in first cluster, second in the second cluster and so on.) Actually this can be done by a number of ways and I have just suggested one simple way for you to begin with. Now apply the algorithm till the centroids (or the means of the clusters ) do not change and in each iteration do what what you have to.
I mean to get a proper answer you will have to be very specific about what problem are u facing in implementation. Answer this or leave a mail.
Happy to help Ujjwal
