You can start with the examples on the documentation page for KMEANS:
You'll need 2 pieces of information. (1) X: your input matrix of features (2) k: number of clusters. Number of clusters is something you may have to start by guessing if you don't already know.
All that is left is to invoke KMEANS:
[IDX,C,sumd] = kmeans(X,k)
If you don't know how many clusters are there in the data, you can make use of cluster evaluation techniques as follows:
