pydata
Add "numerical" function (opposite of C() in formulas)?
It would be nice to have a function, say "N( )", that indicates that input data with pandas dtype "category" should be interpreted as numerical data. In that sense this function would do the exact opposite of C().
Why on earth would I want to do this? Say I have a column 'HoD' (Hour of Day) in the input dataframe whose datatype is a pandas "category" (the data I am working with is huge, so I want to save RAM and not have millions of int64 that take only 24 values). For a simple linear model I just run a vanilla formula specification without using C() and dummy coding is done automatically, that's great. But I also want to run some non-parametric estimators, and for those I want to avoid dummy-coding data. Having such a function would allow to use the same analysis pipeline without having to convert the input data in some cases.
