[LRNRQ] Add Geographically Weighted Random Forest from package spatialML

[G-Lomax]

Algorithm

grf: Geographically Weighted Random Forest Model

Package

spatialML

Supported types

• classif
• clust
• dens
• regr
• surv

I have checked that this is not already implemented in

• mlr3
• mlr3learners
• mlr3extralearners
• Other core packages (e.g. mlr3proba, mlr3keras)

Why do I think this is a useful learner?

Existing implementations of random forest are blind to spatial structure and non-stationarity in data, despite being widely used for modelling geospatial datasets such as satellite remote sensing. Geographically Weighted Regression accounts for spatial non-stationarity using local submodels with a tunable bandwidth parameter.

Further Optional Comments

grf uses the ranger package implementation of RF.