search

DeaglanBartlett / roxy

5 stars 0 forks source link

AIC/BIC comparison for GMM #2

Closed DeaglanBartlett closed 1 year ago

DeaglanBartlett commented 1 year ago

A priori, if using the GMM functionality, one does not know how many Gaussians to use. It would be useful to have a function which runs through ngauss=1,..,N and then computes the AIC and/or BIC for each number. One can then select the best model based on this.

DeaglanBartlett commented 1 year ago

Now implemented in the function find_best_gmm of regressor https://github.com/DeaglanBartlett/roxy/blob/7ed3015c46a4f77a0b4681406c9fdedfac8d300d/roxy/regressor.py#L381