autopls (version 1.3)

postprocessing: Test for model extrapolations or interpolations and removal of bold predictions in autopls

Description

Departures of values predicted by autopls from the original data space and removal of exceedingly extrapolated predictions.

Usage

liability (object, prediction) confine (object, prediction, tolerance)

Arguments

object

object of class autopls

prediction

predicted values as single vector or single layer raster image (RasterLayer from package raster).

tolerance

maximum departure of preserved prediction values

Value

Vector or raster image depending on the type of prediction Uncertainties (liability function) are given in original units. After confine, values exceeding tolerance are replaced by NA.

Details

Takes its time with large images and many objects used in calibration.

See Also

autopls, predict.autopls

Examples

  ## load predictor and response data to the current environment
  data (murnau.X)
  data (murnau.Y)
  
  ## call autopls with the standard options
  model <- autopls (murnau.Y ~ murnau.X)

  ## new data
  new <- murnau.X + 500 

  ## prediction
  pred <- predict (model, new)

  ## check uncertainty
  liability (model, pred)

  ## remove predictions with uncertainty value > 5
  confine (model, pred, 5)