validate: Check the accuracy of a model using independent validation...

In henkelstone/NPEL.Classification: Classifaction and data-handling routines for NPEL Caribou Project

Description

This function is a wrapper for classAcc that, given an independent validation dataset, uses the model to come up with new predicted values. It then calls classAcc with these two validation datasets (the one generated by the model and the true values) to report class accuracies.

Usage

validate(model, valid, ...)

Arguments

model	the model to test.
valid	the validation dataset; must contain all the parameters used in the model.
...	other variables to pass to classAcc.

Value

Returns the result from classAcc

See Also

classAcc

Examples

data ('siteData')
# With categorical data
gen <- sample(1:nrow(siteData),floor(nrow(siteData)*0.5))
modelRun <- generateModels(data = siteData[sort(gen),],
                           modelTypes = suppModels,
                           x = c('brtns','grnns','wetns','dem','slp','asp','hsd'),
                           y = 'ecoType',
                           grouping = ecoGroup[['identity','transform']],
                           echo = FALSE)
valid <- siteData[-gen,]
valid$ecoType <- as.factor(ecoGroup[['identity','transform']][valid$ecoType])
validate(modelRun[[2]],valid)
validModels(modelRun,valid)

# With continuous data
gen <- sample(1:nrow(siteData),floor(nrow(siteData)*0.5))
modelRun <- generateModels(data = siteData[sort(gen),],
                           modelTypes = contModels,
                           x = c('brtns','grnns','wetns','dem','slp','asp','hsd'),
                           y = 'easting',
                           echo = FALSE)
valid <- siteData[-gen,]
validate(modelRun[[2]],valid)
modelsValid(modelRun,valid)

henkelstone/NPEL.Classification documentation built on May 17, 2019, 3:42 p.m.