forestDSS: Construct of a random forest based on training set 'df'.
In sib-swiss/dsSwissKnife: DataSHIELD Tools and Utilities - server side

forestDSS

R Documentation

Construct of a random forest based on training set 'df'.

Description

Calls 'randomForest::randomForest' with formula dep_var ~ expl_vars. The number of trees produced is proportional to the number of rows in the dataset, but with a minimum of 10 and a maximum of 150 trees.

Usage

forestDSS(...)

Arguments

`df:`	[data frame] The training dataset.
`dep_var:`	[string] the response factor ("y"), i.e. the categories that will be the leaves of the tree.
`expl_vars:`	[vector[string]] the classification variables.
`nodesize:`	minimal number of points in the terminal nodes. We fix this high enough to ensure more privacy. Default is 1 for classification and 5 for regression.
`...:`	other parameters to 'randomForest'.