This should merge to master what eventually becomes shapr 1.0.0.
All changes up until that point should be made through separate PRs to be merge into the shapr_1.0.0 branch
Required in this PR
[x] Add adaptive estimation
[x] Add verbose
[ ] Get shaprpy to work again
[ ] Get explain_forecast to work again
[ ] Update explain_forecast to the new adaptive version
[ ] Add asymmetric/causal to explain
Other mandatory TODOs
[ ] Update the NEWS file
Minor TODOs (some might be skipped before release)
[ ] Consider moving compute_sd, n_boot_samps, max_batch_size, min_n_batches, saving_path out of adaptive_arguments and into a new argument list called something like additional_estimation_args
[ ] Consider renaming adaptive_arguments to adaptive_args
[ ] Consider moving keep_samp_for_vS and MSEv_uniform_comb_weights into a input arg called something like output_args
[ ] Consider renaming shapley_reweighting to kernelSHAP_reweighting
[ ] A bit more cleaning on all vignettes (all are currently without code errors, and at an OK level, I think)
[ ] A bit more cleaning on the documentation (also OK at this stage I think, but new eyes don't hurt)
[ ] Delete unnecessary scripts and devel code and other leftover from inst
[ ] Consider adding parallelization to the boostrapping
[ ] Consider moving saving_path and iter_res_dt to the main output to make it easier to access for users
martinju
commented
1 month ago
This should merge to master what eventually becomes shapr 1.0.0. All changes up until that point should be made through separate PRs to be merge into the shapr_1.0.0 branch
Required in this PR
Other mandatory TODOs
Minor TODOs (some might be skipped before release)
additional_estimation_argsadaptive_argumentstoadaptive_args