Visualising 2Way Random Forest Interactions
Utilize random forest variable importance and contribution in order to visualise interactive 3d surfaces of interaction effects. The approach is that of data subsampling and smoothing, in order to experiment with 2-way effect visualizations and to accomodate results from larger data sets that are more difficult to be viewed.
You can install the package with:
devtools::install.packages("vis2rfi")
See the package’s website for usage and examples.