Lecture46 (Data2Decision) Best Subset Regression in R

[@JoelleRocks03]

Thank you so much for making this. You have no idea how helpful it was!

[@multitaskprueba1]

Fantastic video! Thank you so much!

[@taximxitxi]

I'm always getting this error as soon as I run my "leaps" --> Error in `contrasts

[@AshPhoenix]

"Reorder variables and try again:" how should I handle this? Thanks,

[@rohittawde5105]

When we consider the interactions within the six independent features the total number of features would now be close to twenty seven. However in the cp vs subset size plot we observe that it subset size is only until nine or ten. Is there an upper bound on the subset size that leaps package considers? Thanking you in anticipation.

[@kalyanasundaramsp8267]

brilliant sir..great....sir, when you have discrete variables instead of leaps what should we use?....

[@bevansmith3210]

Hi Chris, excellent video thanks. Once we have our predictors, how do we find their beta coefficients and p-values? Thanks again.

[@ChrisMack]

You can just run a linear regression. See this lecture: kzbin.info/www/bejne/fpWoXpJnf5iUh5Y

[@bevansmith3210]

Ah, of course. Thanks again.

[@ycdantywong]

How do I extract the variables names from the subset function ? Just a plot itself isn't all that useful.......

[@kalyanasundaramsp8267]

when we discrete variables what function we should use instead of leaps...

[@arnabaich3132]

subset plot method is good for single digit no of co-variates. I have 12 covariates. I couldn't figure out anything from the plots since it is too messed up .I think for large no of covariates we have to rely on the summary of regsubset method and checek the variable matrix.

[@mehdinaderi7241]

Thanks, very informative

[@andychoo9605]

thanks for saving my homework

[@oscarhuang8458]

thank you sir!