postprocessing: Test for model extrapolations or interpolations and removal...
In autopls: Partial Least Squares Regression with Backward Selection of Predictors

Description Usage Arguments Details Value Author(s) See Also Examples

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

1 2	liability (object, prediction) confine (object, prediction, tolerance)

`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

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

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.

Sebastian Schmidtlein

autopls, predict.autopls

  ## 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)