buildBalancedModelData: Build Balanced Model Data

In adamkc/DIMEChelper: Companion functions for DIMEC

Description

This function takes a starting folder "Original" and splits the classes within that folder into Training, Validation, and Testing data. The data is balanced so that the smallestclass dictates how much training data is added for each class. The remainders are added to the validation dataset.

Usage

buildBalancedModelData(data.folder, trainpercent = 0.8,
  validatepercent = 0.1, testpercent = 0.1)

Arguments

data.folder	Top directory that contains "Original" folder. "Model Data" will be added to this directory as well.
trainpercent	Percent of data to add to the training dataset. These three need to add to 1.
validatepercent	Percent of data to add to the validation dataset. These three need to add to 1.
testpercent	Percent of data to add to the test dataset. These three need to add to 1.

Value

Builds a balanced dataset from the Original folder. To be replaced eventually by buildModelData with balancing logical argument.

adamkc/DIMEChelper documentation built on Feb. 1, 2020, 12:24 p.m.