plot.gspcrcv: Plot the cross-validation solution path for the GSPCR...
In gspcr: Generalized Supervised Principal Component Regression

plot.gspcrcv

R Documentation

Plot the cross-validation solution path for the GSPCR algorithm

Description

Produces a scatter plot showing the CV score obtained by cv_gspcr (Y-axis) with different threshold values (X-axis) for a different number of components (lines).

Usage

## S3 method for class 'gspcrcv'
plot(
  x,
  y = NULL,
  labels = TRUE,
  errorBars = FALSE,
  discretize = TRUE,
  y_reverse = FALSE,
  print = TRUE,
  ...
)

Arguments

`x`	An object of class `gspcr`.
`y`	The CV fit measure to report on the Y axis. Default is the fit measure specified in `cv_gspcr()`.
`labels`	Logical value. `FALSE` hides the labels of the points indicating the number of components used. The default is `TRUE`.
`errorBars`	Logical value. `TRUE` shows the error bars for each point. The default is `FALSE`.
`discretize`	Logical value. `TRUE` treats the X-axis as a discrete measure that facilitates comparing solution paths between different fit measures. The default is `TRUE`.
`y_reverse`	Logical value. `TRUE` reverses the y axis scale. The default is `FALSE`.
`print`	Logical value. TRUE prints the plot when the function is called. The default is `TRUE`.
`...`	Other arguments passed on to methods. Not currently used.

Details

The bounds defining the error bars are computed by cv_gspcr(). First, the K-fold cross-validation score of the statistic of interest (e.g., the F score, the MSE) is computed. Then, the standard deviation of the statistic across the K folds is computed. Finally, the bounds used for the error bars are computed by summing and subtracting this standard deviation to and from the K-fold cross-validation score of the statistic.

Reversing the y-axis with y_reverse can be helpful to compare results obtained by different fit measures.