classifying data to caculate transition probabilities

 Hi all
I have a 23×36 matrix of data which each column is a time series of a random variable having Normal distribution. I want to classify each column of data into subclasses in such way that the probability of data of classes be equal.

Could you please guide me.

Thanks

Answer 1

[Walter Roberson]

In order for the frequency of each group to be equal when the total number of data elements is 23, then either you need one single group of size 23, or else 23 groups each of size 1. Any other arrangement will end up with at least one group having a different frequency than the other groups.

Answer 2

[Daniel Shub]

This seems like a pretty straight forward classification problem. What have your tried so far? Assuming your data satisfy a few constraints (Normal gets you most of the way there) a perceptron should work (or you could use glmfit or correlation).