tidymodels / recipes

Pipeable steps for feature engineering and data preprocessing to prepare for modeling
https://recipes.tidymodels.org
Other
562 stars 111 forks source link

Random projection step #172

Open alexpghayes opened 6 years ago

alexpghayes commented 6 years ago

I'd like a signature along the lines of:

step_randomproj(recipe, ..., dimensions, error)

where the user can either specify the number of dimensions to project into or an acceptable level of error in the projection (then we could back out dimension from there).

It would also be nice be able to project into an arbitrary number of dimensions (i.e. also higher dimensions than the original data) to facilitate working with extreme learning machines or other random feature methods.

alexpghayes commented 5 years ago

Reference: https://jeremykun.com/2016/02/08/big-dimensions-and-what-you-can-do-about-it/