A MLJ wrapper to the R HAL package.
The project relies on RCall, if you don't want to fallback on a Conda installation, you should make sure the R_HOME
environment variable is correctly set. For known issues regarding RCall compatibility look into the RCall intallation section.
You will also need to install the original HAL package in your R environment.
This wrapper can then be installed via:
add HighlyAdaptiveLasso
using HighlyAdaptiveLasso, MLJ, Random
X = randn(100, 4)
y = randn(100)
hal = HALRegressor()
mach = machine(hal, X, y)
evaluate!(mach)