“Because of performances it was decided that Random Forest classifier should be implemented in Python (it shows better performances than random forest package in R).” Given that the author used a standard sklearn RandomForestClassifier, I’d be surprised if the ranger implementation in R was the one that was tested in whatever benchmark he or she read. I ran both versions last spring, GitHub - ercbk/nested-cross-validation-comparison: Experimenting with various implementations and methods of nested cross-validation in R and Python, and sklearn’s rf was very slow.