Skip to contents

Plot a bgam fit

Source: R/print.R
plot.bgam.Rd

Produces a 4-panel diagnostic display for a fitted component-wise P-spline boosting model:

Usage

# S3 method for class 'bgam'
plot(x, ...)

Arguments

x

An object of class "bgam", as returned by bgam().

...

Currently ignored.

Value

Invisibly returns x.

Details

  1. Partial effect curve \(\hat{f}_{j^*}(x_{j^*})\) for the top predictor by selection frequency, plotted over a fine grid covering the training range.

  2. Selection-frequency bar chart for all predictors (capped at 20 bars), sorted descending. Use this panel to identify which predictors drive the model.

  3. Training loss path (MSE for gaussian, mean deviance per observation for binomial) vs boosting iteration. A red dashed vertical line marks mstop_opt when autotune = TRUE.

  4. Fitted vs Observed (gaussian) or fitted-probability box plots by class (binomial).

The function works in headless (non-interactive) mode without error.

References

Buehlmann, P. and Yu, B. (2003). Boosting with the L2 loss: Regression and classification. Journal of the American Statistical Association, 98(462):324–339. doi:10.1198/016214503000125

Hofner, B., Mayr, A., Robinzonov, N. and Schmid, M. (2014). Model-based boosting in R: A hands-on tutorial using the R package mboost. Computational Statistics, 29(1–2):3–35. doi:10.1007/s00180-012-0382-5

See also

bgam(), print.bgam(), summary.bgam()

Examples

# \donttest{
set.seed(1)
n <- 200
x <- matrix(rnorm(n * 4), n, 4)
y <- sin(x[, 1]) + 0.5 * x[, 2]^2 + rnorm(n, sd = 0.5)
fit <- bgam(x, y, mstop = 50, autotune = FALSE)
plot(fit)

# }