Todos: - Githubissues

karink520 commented 3 years ago

If you want to tackle any of these, just reply below, and I'll mark that you're on it!

[x] additional test coverage - Gabe and Karin
[x] legend for color in precinct_level_plot Gabe
[x] show examples of how to set sampling parameters (number of draws, tuning samples, etc....) in notebooks Karin
[x] add more exposition in notebooks to make them more beginner-friendly - Karin
[x] truncated normal - Karin (implemented a version with pymc - can also add one that's closer to king97)
[x] add plots and reporting for differences between racial groups (done for 2 x 2, need r x c) Gabe
[x] add candidate of interest reporting Amy
[x] Make rbyc model take precinct_names again Karin
[x] Generalize plot_precinct_scatterplot to work with 2x2 Karin
[x] Prettify plots Gabe
[x] Q& G - Karin
[ ] re-enable alerts for wakefield method Karin
[x] Compare to eiPack Gabe, Amy, Karin
[x] revise JOSS draft - especially need revisions/feedback on discussion of existing R packages https://github.com/mggg/ecological-inference/tree/paper/paper
[ ] Make precinct-level kdes plottable by candidate OR group
[x] Add example of r by c precinct-level plots to notebook(s)
[x] Add example of histogram option of precinct-level plot to notebook(s)

[x] Generalize 2x2 plotting functions plot_precincts, plot_conf_or_credible_interval to RxC Gabe
[x] Try on Texas data, maybe w/ pickling Gabe

gabeschoenbach commented 3 years ago

I can also take the color legend in precinct_level_plot and adding plots / reporting differences between racial groups for RxC

karink520 commented 3 years ago

Amazing! TwoByTwoEIBaseBayes.polarization_report(), TwoByTwoEIBaseBayes._calculate_polarization(), TwoByTwoEIBaseBayes.plot_polarization_kde() what's implemented in the 2x2 case (that last one uses plot_utils.plot_polarization_kde() ), added and merged today, with examples uses in santa_clara_demo.ipynb (look for the green kdes).

gabeschoenbach commented 3 years ago

FYI, added a TODO about generalizing the plotting functions

bblpo commented 3 years ago

Great contribution to many fields. Thanks to your hard work.

One thing that may be more useful to do is to make covariates an option for 2x2. See the last example from the original post: https://gist.github.com/ColCarroll/9fb0e6714dc0369acf6549cededcc875

mggg / ecological-inference

Todos: #35