Julia for Data Science

Implementation in Julia

Random forests are available in the Julia-registered packages from Kenta Sato:

Pkg.update() Pkg.add("RandomForests") 

This is a CART-based random forest implementation in Julia. This package supports:

• Classification models
• Regression models
• Out-of-bag (OOB) errors
• Feature importances
• Various configurable parameters

There are two separate models available in this package:

• Classification
• Regression

Each model has its own constructor that is trained by applying the fit method. We can configure these constructors with some keyword arguments listed as follows:

RandomForestClassifier(;n_estimators::Int=10, 
                        max_features::Union(Integer, FloatingPoint, Symbol)=:sqrt, 
                        max_depth=nothing, 
                        min_samples_split::Int=2, 
                        criterion::Symbol=:gini)