search

stephenslab / ebnm

R package to fit Empirical Bayes Normal Means model.
https://stephenslab.github.io/ebnm/
12 stars 9 forks source link

Interface changes, optimization improvements, more point-Laplace functionality #18

Closed willwerscheid closed 5 years ago

willwerscheid commented 5 years ago

Includes most changes suggested in #16 (still need to do npmle). Replaces calls to optim with nlm, which uses the Hessian and makes more sense for problems of this size. Implements a bunch of things for ebnm_point_laplace, including fixing the scale, calculating lfsr, handling infinite SEs, and sampling from the posterior. Fixes #11 .

codecov-io commented 5 years ago

Codecov Report

Merging #18 into master will increase coverage by 26.19%. The diff coverage is 90.66%.

Impacted file tree graph

@@             Coverage Diff             @@
##           master      #18       +/-   ##
===========================================
+ Coverage    63.1%   89.29%   +26.19%     
===========================================
  Files          14       20        +6     
  Lines         458      682      +224     
===========================================
+ Hits          289      609      +320     
+ Misses        169       73       -96
Impacted Files Coverage Δ
R/mle_pn_special_cases.R 59.37% <0%> (+9.37%) :arrow_up:
R/nlm_fn_point_laplace.R 100% <100%> (ø)
R/summary_results_point_normal.R 96.77% <100%> (ø) :arrow_up:
R/nlm_fn_point_normal.R 100% <100%> (ø)
R/output.R 100% <100%> (ø) :arrow_up:
R/ebnm_normal_scale_mixture.R 100% <100%> (ø)
R/ebnm_unimodal.R 100% <100%> (ø)
R/post_sampler_point_normal.R 100% <100%> (ø) :arrow_up:
R/ebnm_normal.R 100% <100%> (ø)
R/mle_point_normal.R 80% <64.51%> (-0.44%) :arrow_down:
... and 20 more

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 3dbfd89...0313ff1. Read the comment docs.