Closed ejolly closed 6 years ago
This is now part of #149
Left you feedback on the actual code. I'm on the fence about whether it makes sense to have a single regression method or if we should multiple ones that can be used in the same way. For example, predict is a method that allows you to use many different algorithms. It's advantage is that it is easy to use. the disadvantage is that passing **kwargs can be a bit opaque.
fixed with PR #149
Working but slow implementation for a Brain_Data method that fits ARMA model to voxel responses rather than standard OLS. It operates very much like AFNI's 3dREMLfit but using MLE via Kalman filter instead of REML.