Time-series from real dataset

[fpgdesa]

Hi! I have a multidimensional time series of which i select a few features (just 1 feature, for example). I want to determine how many states the mcmc identifies using N samples. I have some questions about the best way for input data into MCMC processing. Let X be an array : X = [ x_1, x_2, ..., x_N]. Would I need to perform which transformation in these data before call the function MCMC? Until now, the function return just 1 state after 500 iteration (as the example).

Best regards,

Thank you!