best_kmeans(X)

This function can determine the best cluster numbers in clustering using k-means method.

[IDX,C,SUMD,K] = best_kmeans(X) partitions the points in the N-by-P data matrix X
into K clusters. Rows of X correspond to points, columns correspond to variables.
IDX containing the cluster indices of each point.
C is the K cluster centroids locations in the K-by-P matrix C.
SUMD are sums of point-to-centroid distances in the 1-by-K vector.
K is the number of cluster centriods determined using ELBOW method.
ELBOW method: computing the destortions under different cluster number counting from
1 to n, and K is the cluster number corresponding 90% percentage of
variance expained, which is the ratio of the between-group variance to
the total variance. see <http://en.wikipedia.org/wiki/Determining_the_number_of_clusters_in_a_data_set>
After find the best K clusters, IDX,C,SUMD are determined using kmeans
function in matlab.