Reimplement and expose weighted estimator

CoryMcCartan / birdie

Bayesian Instrumental Regression for Disparity Estimation

http://corymccartan.com/birdie/

GNU General Public License v3.0

5 stars 3 forks source link

Reimplement and expose weighted estimator #9

Closed CoryMcCartan closed 1 year ago

CoryMcCartan commented 1 year ago

Perhaps OLS as well.

[x] Names: est_wtd(), est_ols() ?
[x] Class: maybe move common functions (coef(), plot(), fitted()) etc to a new S3 class, and then have birdie-specific functionality in generics for that class only. birdie() return value would have class c("birdie", "est_model") or something, while wtd and OLS would have latter only.
[x] Implementation: use dirichlet_map()
[x] Documentation

CoryMcCartan commented 1 year ago

Or maybe have it the other way: make est_*() inherit from birdie() (minimal code changes, no one really sees this anyway) and just implement a new summary() and print() on top. As long as list format is the same all the existing generics should carry over fine.