rsMove (version 0.2.1)

moveModel: moveModel

Description

Spatially stratified predictive modeling.

Usage

moveModel(p.data = p.data, a.data = a.data, label = NULL, fun = fun,
  nruns = 1)

Arguments

p.data

Object of class data.frame with environmental variables for presence samples.

a.data

Object of class data.frame with environmental variables for background samples.

label

Region labels. If missing, "p.data" is assumed as one region.

fun

A function with the modeling algorithm to use.

nruns

Number of runs. Default is 1.

Value

A list.

Details

For n iterations, where n is the number of unique sample regions, the function uses one sample region for validation while the remaining ones are used for training. The background samples are split randomly at each iteration. The final accuracy, provided as a F1-score for both presence and background sampels, is derived from the total of true and false positives ($f1). Additionaly, for each run, the function returns a model ($model) which is trained using all the samples. This output can be passed to modelApply(). By default, the function uses a Random Forest classifier. However, the a user specified function can be passed using fun.

See Also

sampleMove labelSample backSample modelApply train

Examples

# NOT RUN {
 
 require(rgdal)
 require(raster)
 require(sp)
 
 # read remote sensing data
 file <- list.files(system.file('extdata', '', package="rsMove"), 'tc.*tif', full.names=TRUE)
 rsStk <- stack(file)
 
 # read movement data
 moveData <- read.csv(system.file('extdata', 'konstanz_20130805-20130811.csv', package="rsMove"))
 moveData <- SpatialPointsDataFrame(moveData[,1:2], moveData, proj4string=crs(rsStk))

 # retrieve remote sensing data for samples
 rsQuery <- dataQuery(xy=moveData,img=rsStk, remove.dup=TRUE)
 
 # identify unique sample regions
 label <- labelSample(xy=rsQuery, rad=3000, pxr=rsStk)
 
 # select background samples
 ind <- which(label>0) # selected samples
 bSamples <- backSample(xy=moveData[ind,], rid=label[ind], img=rsStk, method='pca')
 
 # derive model predictions
 fun <- function(x,y) {train(x, y, method="rf", trControl=trainControl(method='oob'))}
 out <- moveModel(p.data=rsQuery@data, a.data=bSamples@data, label=label, fun=fun, nruns=1)

# }