Description of the problem/new feature
Since beads and jbcd produce global fits of the baseline whose curvature is controlled by the parameters alpha and beta, respectively, then optimize_extended_range should theoretically work well at selecting the optimal value for them. The plots below show the output baseline and 'rmse' param values using a patched version of optimize_extended_range with beads and jbcd, which demonstrates that this is indeed the case.
This should wait until the optimize_lam branch is merged since optimizers also need a rework on how they filter which algorithms are allowed to be used with them. The filtering should not be based on modules as it currently is, but instead should just look at if the signature of the selected method has required keywords (eg. collab_pls would look for weights, optimize_extended_range would look for poly_order, lam, alpha, beta) with some additional exclusions/special-cases accounted for.
Description of the problem/new feature
Since
beadsandjbcdproduce global fits of the baseline whose curvature is controlled by the parametersalphaandbeta, respectively, thenoptimize_extended_rangeshould theoretically work well at selecting the optimal value for them. The plots below show the output baseline and 'rmse' param values using a patched version ofoptimize_extended_rangewithbeadsandjbcd, which demonstrates that this is indeed the case.This should wait until the
optimize_lambranch is merged since optimizers also need a rework on how they filter which algorithms are allowed to be used with them. The filtering should not be based on modules as it currently is, but instead should just look at if the signature of the selected method has required keywords (eg.collab_plswould look forweights,optimize_extended_rangewould look forpoly_order,lam,alpha,beta) with some additional exclusions/special-cases accounted for.