[R] stepAIC formula upper limit guidelines

[Uwe Ligges]

Tate Avery wrote:
> Hello,
> 
> I am attempting to refine an lm()-generated model using the stepAIC 
> function.
> 
> My model has approximately 20 inputs and I am trying to determine the best 
> upper limit scope for using those inputs.
> 
> My lower limit is "y ~ 1" and my original upper limit was "y ~ x1 + x2 + ... 
> + x20".
> 
> This is great start, but I am wondering if some other (more broad) upper 
> limit would allow for combinations of inputs (examples: "x1*x3" or "x5/x7") 
> before the coefficients are applied.


Are you really sure you want to try out all those combinations of 
variables? You will need an almost infinite number of observation to be 
sure you did not find anything just by chance ...

Uwe Ligges



> Specifically, I am trying to determine what syntax will give stepAIC the 
> most flexibility to slice and dice my inputs as it searches for the best 
> possible model.
> 
> Any details or references would be immensely appreciated.
> 
> Thank you,
> Tate Avery
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html