Methods returning data.frame of decision criteria values such as RSquare, Adjr, AIC etc must also return the equivalent regression model object on-demand.

[linusjf]

https://github.com/rsquaredacademy/olsrr/issues/158#issuecomment-623302710

Brief description of the problem

Methods that return the model parameters and it's decision and information criteria such as RSquare, Adjr, Aic etc must also return the regression model object (class lm) on-demand. The lightweight object (excluding the regression model objects specified above) must include the response variable name (this is currently missing).

The mindex column of the data.frame indexes into the list of models returned. The list can be a named list with the names a combination of the predictors in the model. The named identifier could be a concatenation of the predictors' names sorted alphabetically.

[jens-daniel-mueller]

Hi @aravindhebbali,

the propose return of model coefficients would be very helpful for my work. Are you planning to implement this, and maybe already have an idea when it might become available?

In ols-all-possible-regression.R from line 281 ff it seems that you started working on it. Is this correct? Can you give a quick status update?

I would also be happy to help, but would probably require instructions as I'm not an experienced contributor.

Thanks for your feedback. Jens

[aravindhebbali]

Hi @jens-daniel-mueller,

We are planning a release of olsrr by November end. I will update you about the API changes and how you can contribute to it in a couple of days.

Thanks & Keep Safe, Aravind Hebbali