predict.wa {analogue}R Documentation

Predict from a weighted average model

Description

Model predictions and cross-validation predictions for weighted averaging transfer function models.

Usage

## S3 method for class 'wa':
predict(object, newdata,
        CV = c("none", "LOO", "bootstrap", "nfold"),
        verbose = FALSE, n.boot = 100, nfold = 5, ...)

Arguments

object an object of class "wa", usually the result of a call to wa
newdata An optional data frame in which to look for variables with which to predict. If omitted, the fitted values are used.
CV Should cross-validation be performed? Leave-one-out ("LOO"), bootstrap ("bootstrap") and k-fold ("nfold") CV are currently available.
verbose Should CV progress be printed to the console?
n.boot The number of bootstrap samples or k-fold steps.
nfold Number of subsets in k-fold CV.
... further arguments passed to or from other methods.

Details

Not all CV methods produce the same output. CV = "bootstrap" and CV = "nfold" produce sample specific errors.

Value

An object of class "predict.wa", a list with the following components:

pred A list with components pred and rmsep containing the predicted values and the sample specific errors if available.
performance A list with model performance statistics.
model.pred A list with components pred and rmsep containing the predicted values for the training set samples and the sample specific errors if available.
call the matched function call.
CV.method The CV method used.

Author(s)

Gavin L. Simpson and Jari Oksanen (k-fold CV)

References

Birks, H.J.B., Line, J.M., Juggins, S., Stevenson, A.C. and ter Braak, C.J.F. (1990). Diatoms and pH reconstruction. Philosophical Transactions of the Royal Society of London; Series B, 327; 263–278.

See Also

wa, predict.mat, performance, reconPlot.

Examples

## continue the example from ?wa
example(wa)

## load the RLGH data
data(rlgh)
rlgh <- rlgh / 100

## Predict pH for the RLGH samples
rlgh.pred <- predict(mod, rlgh, CV = "bootstrap", n.boot = 100)

## draw the fitted reconstruction
reconPlot(rlgh.pred, use.labels = TRUE, display = "bars")

## extract the model performance stats
performance(rlgh.pred)

[Package analogue version 0.6-6 Index]