4 Comments
User's avatar
Cynthia F ORourke's avatar

My introduction to autoML was MuMIn, an R package that didn't pitch itself as autoML, but rather pitched [generalized] linear model selection as an information theoretics problem. AutoML at its best, for me, remains an algorithm-agnostic solution to an information theoretics problem.

Expand full comment
Thomas W. Dinsmore's avatar

Found here, for anyone who wants to check it out

https://cran.r-project.org/web/packages/MuMIn/index.html

Expand full comment
Gonzo's avatar

I recapitulated this history in grad school down to most of the tools—though i did use JMP for a semester before returning to SAS and I didn’t write any fortran until I started using R.

When i first learned neuralnets it was 2002 and my stats professor cautioned me: Neural networks are what really smart people who believe magic end up sacrificing their careers for. After a summer dalliance I got back to work with glm and some agent based modeling (ecology).

Don’t be them, he said. Alas, I wasn’t them.

Expand full comment
Thomas W. Dinsmore's avatar

Thanks for reading and commenting!

Expand full comment