I would like to use rpy2 to call the regsubsets function from the R leaps package. Sine there seems to be no python equivalent (at least non-that I have found).
import pandas
import rpy2.robjects as robjects
from rpy2.robjects import pandas2ri
import rpy2.robjects.packages as rpackages
pandas2ri.activate()
leaps = rpackages.importr("leaps")
x = pandas.DataFrame(np.random.randn(10,10))
y = pandas.DataFrame(np.random.randn(10))
leaps.regsubsets(x=x, y=y, nbest=10, nvmax=3, really.big=T)
Calling that function does require the argument really.big=T. That of course does not work, because in python the . has a direct function as opposed to R. (i get a "SyntaxError: keyword can't be an expression"). Can anyone tell me how to call it from rpy2?
I have tried to do it with kwargs like leaps.regsubsets({"x":x ... "really.big":T}) but that also does not work and I also do not know what "T" is in R.
OK, I found out that the "." gets transleted to "_". but sadly it still does not work, somehow the argument does not get passed on.
leaps.regsubsets(x=X, y=Y, nbest=10, nvmax=3, really_big=True)
still gives me the error: RRuntimeError:
Error in leaps.exhaustive(a, really.big = really.big) :
Exhaustive search will be S L O W, must specify really.big=T
The documentation link in the comments in for rpy2-2.2.x that is quite old. Documentation for the current release is at:
https://rpy2.github.io/doc/v3.0.x/html/robjects_functions.html
Otherwise the documentation is rather clear about the near-impossibility of extracting parameters that can be accepted in the R ellipsis (...) (which is what @CristiFati is pointing out in the comments).
You can used syntactically-invalid parameter names with **. For example here:
leaps.regsubsets(x=x, y=y,
nbest=10, nvmax=3,
**{'really.big': true})