TODO.md

In Netflix/sherlock: Causal Machine Learning for Segment Discovery and Analysis

TO-DO

16 Aug 2021

• [x] fix rounding bug in cost_knapsack for better integer approximations

12 Aug 2021

• add documentation in all R source files for estimation functionality:
• [x] cate.R
• [ ] data.R
• [x] effects.R
• [x] fluctuation.R
• [x] nuisance.R
• [x] costs.R
• [x] segment.R
• [x] confint.R
• [x] setup_data.R
• [x] wrappers.R
• [x] Lrnr_prob_known.R
• add documentation where necessary in all R auxiliary source files:
• [x] bound.R
• [x] scale.R
• [x] globals.R
• [x] options.R
• [x] zzz.R

11 Aug 2021

• [x] fix TMLE fluctuation to fit each parameter separately
• [x] add confidence intervals for CATE and effect measure estimates
• [ ] add in checks for diagnostics at smaller sample sizes

07 Aug 2021

• [x] reduce to a few high-level functions: sherlock to detect segments, watson to assign an optimal rule, and mycroft to compute effects

06 Aug 2021

• [x] rename HTE to OTE, since HTE already defined otherwise in ABlaze
• [ ] evaluate OTE conditional on segmentation covariates, ie, COTE = E(OTE|V)

04 Aug 2021

-[x] create a wrapper function to bundle set_est_data and est_cate -[x] add summarization of EIF across IDs for repeated measures -[x] add functionality to handle known g in PS estimation -[x] export global option for sl3.pcontinuous to avoid guessing categorical

Netflix/sherlock documentation built on Dec. 17, 2021, 5:22 a.m.