search

radiant-rstats / radiant.model

Code for the Model menu in Radiant
http://radiant-rstats.github.io/radiant.model
Other
19 stars 16 forks source link

Added robust standard errors #1

Closed bjornerstedt closed 6 years ago

bjornerstedt commented 7 years ago

Robust standard errors in OLS are important in econometrics, even at the introductory level. Here is a basic implementation for radiant.model::regress.

/Jonas

vnijs commented 7 years ago

Thanks @bjornerstedt! I'll review soon and make sure this feature is added. Is there a particular reason you want to use HC1. I use HC0 for radiant.model::logistic

bjornerstedt commented 7 years ago

Nothing big: I am currently using Stata in my teaching, and it has HC1 as default in robust estimation.

/Jonas

2017-05-19 23:58 GMT+02:00 Vincent Nijs notifications@github.com:

Thanks @bjornerstedt https://github.com/bjornerstedt! I'll review soon and make sure this feature is added. Is there a particular reason you want to use HC1. I use HC0 for radiant.model::logistic

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/radiant-rstats/radiant.model/pull/1#issuecomment-302821000, or mute the thread https://github.com/notifications/unsubscribe-auth/AMA0z4zvhekZBLIm8etQdlM7eXEw2eB8ks5r7hCjgaJpZM4NgtXx .

vnijs commented 7 years ago

I added robust estimation to the github repo using HC1 (incl. coefficient confidence intervals for Summary and Coefficient plot). Initial testing looks good but would appreciate your reaction. Install using:

devtools::install_github("radiant-rstats/radiant.data")
devtools::install_github("radiant-rstats/radiant.basics")
devtools::install_github("radiant-rstats/radiant.model")
vnijs commented 7 years ago

You can get the latest version of radiant.model with the "robust" option for linear and logistic regression by using radiant::update_radiant(). Like I said, initial testing looks good but would appreciate your reaction

vnijs commented 6 years ago

@bjornerstedt I'll close this now. If you have any issues with the current implementation please do let me know.