Apologies for another big one! Need to get better about focusing on one task I guess...
-Moved contrast-model specific code out of nems and into nems_db
-Reorganized xforms module according to function roles
-Added several general-purpose keywords related to fitting, initialization and modelspec manipulation that are used by the contrast model but may be useful for others.
-Updated fit_from_priors analysis
-Corrected a bug in nems.analysis.test_prediction that was resulting in a list of identical recording objects despite evaluating on different modelspecs.
-Added correlation_per_model function to test_prediction for saving correlation information on a per-modelspec basis.
-Added get_best_modelspec function to nems.modelspecs for forcing only the "best" modelspec to be kept after fitting (useful when fitting from randomly sampled initial conditions). Intend to improve this in the future to sort the modelspecs instead so that the other fitted modelspecs are not thrown away.
Apologies for another big one! Need to get better about focusing on one task I guess...
-Moved contrast-model specific code out of nems and into nems_db -Reorganized xforms module according to function roles -Added several general-purpose keywords related to fitting, initialization and modelspec manipulation that are used by the contrast model but may be useful for others. -Updated fit_from_priors analysis -Corrected a bug in nems.analysis.test_prediction that was resulting in a list of identical recording objects despite evaluating on different modelspecs. -Added correlation_per_model function to test_prediction for saving correlation information on a per-modelspec basis. -Added get_best_modelspec function to nems.modelspecs for forcing only the "best" modelspec to be kept after fitting (useful when fitting from randomly sampled initial conditions). Intend to improve this in the future to sort the modelspecs instead so that the other fitted modelspecs are not thrown away.