search

rsquaredacademy / olsrr

Tools for developing OLS regression models
https://olsrr.rsquaredacademy.com/
Other
102 stars 22 forks source link

Best subset regression options #157

Closed aravindhebbali closed 4 years ago

aravindhebbali commented 4 years ago

Allow the user to select the criterion in ols_step_best_subset(). Currently, it is R2 but the end user cannot specify the criterion.

linusjf commented 4 years ago

Can the user specify multiple criteria with the result being the union of the criteria specified? Can a user also specify an intersection of multiple criteria?

aravindhebbali commented 4 years ago

Yes.. an option can be given to the user to specify multiple criteria. When you say the union of criteria, if the criteria(s) return model with different set of predictors, the model returned by the function should use the union of those predictors right?

linusjf commented 4 years ago

No, the union of best models for each criteria specified.

What does the existing functionality provide?

Additionally, you can add an option to provide a number of best models, not just one.

linusjf commented 4 years ago

Yes.. an option can be given to the user to specify multiple criteria. When you say the union of criteria, if the criteria(s) return model with different set of predictors, the model returned by the function should use the union of those predictors right?

I don't see why you're asking me this question. It's quite obvious that a union or intersection applies to sets, not to predictors. I've opened a fresh issue --- if that's the issue instead.