partition: Partition raster data

Description

Divide a categorical raster map into training and testing partitions. A wrapper function for caret::createDataPartition (Kuhn, 2008) to divide a categorical raster map into training and testing partitions.

Usage

partition(x, size = 0.5, spatial = TRUE, ...)

Arguments

RasterLayer with categorical data

size

numeric value between zero and one indicating the proportion of non-NA cells that should be included in the training partition. Default is 0.5, which results in equally sized partitions

spatial

logical. If TRUE, the function returns a SpatialPoints object with the coordinates of cells in each partition. If FALSE, the cell numbers are returned

…

additional arguments (none)

Value

A list containing the following components:

train: a SpatialPoints object or numeric vector indicating the cells in the training partition
test: a SpatialPoints object or numeric vector indicating the cells in the testing partition
all: a SpatialPoints object or numeric vector indicating all non-NA cells in the study region

References

Kuhn, M. (2008). Building predictive models in R using the caret package. Journal of Statistical Software, 28(5), 1-26.

Examples

Run this code

# NOT RUN {
# }
# NOT RUN {
## Plum Island Ecosystems

## Load observed land use maps
obs <- ObsLulcRasterStack(x=pie,
                   pattern="lu",
                   categories=c(1,2,3),
                   labels=c("forest","built","other"),
                   t=c(0,6,14))

## create equally sized training and testing partitions
part <- partition(x=obs[[1]], size=0.1, spatial=FALSE)
names(part)

# }