deforestable (version 3.1.2)

createDataPartition: Data Partitioning

Description

As input data, the functions need two folders- Nonforestdir with images of non-forest and forestdir with ones of forest. createDataPartition() splits data into training and testing partitions while keeping the relative sample size of the classes the same as in the original data. createFolds() splits the data into k folds for cross-validation.

Usage

createDataPartition(forestdir, Nonforestdir, times = 1, p = 0.5)
createFolds(forestdir, Nonforestdir, k = 5)

Value

createDataPartition returns a list of data partitions. Each partition consists of 4 sets- forest training, non-forest training, forest test and non-forest test set. createFolds returns lists $forest and $nonforest with k folds in each of them.

Arguments

forestdir

path to the directory with (only) forest images

Nonforestdir

path to the directory with (only) non-forest images

times

the number of data partitions to make

p

the percentage of data to set aside for training

k

the number of folds to split the data into

Functions

• createFolds(): Split data into folds

Examples

library(deforestable)
forestdir <- system.file('extdata/Forest/', package = "deforestable")
Nonforestdir <- system.file('extdata/Non-forest/', package = "deforestable")

trainPart <- createDataPartition(forestdir=forestdir, Nonforestdir=Nonforestdir, p = .7, times = 1)

folds <- createFolds(forestdir, Nonforestdir, k = 10)