cgaim (version 1.0.3)

plot.cgaim: Plot ridge function

Description

Plot method for the ridge and smooth terms of a cgaim object. If provided, also plots confidence intervals.

Usage

# S3 method for cgaim
plot(x, select = NULL, ci = NULL, ci.plot = c("polygon",
  "lines"), ci.args = list(), add = FALSE, xcenter = FALSE,
  xscale = FALSE, yshift = FALSE, yscale = FALSE, ...)

Value

The function is called to generate plots and returns no value.

Arguments

x

A cgaim object.

select

A numeric or character vector indicating which terms to plot.

ci

An object returned by a call to confint.cgaim. If NULL, no confidence interval is drawn.

ci.plot

Whether to plot the confidence intervals as shaded areas ci.plot = "polygon" or as lines ci.plot = "lines".

ci.args

Additional arguments to be passed to the function used to draw confidence interval. Either polygon or lines.

add

Logical. If TRUE, adds the function to the current active plot.

xcenter, xscale

Centering and scaling values for the x axis. See scale.

yshift, yscale

Either logical or numeric values to shift and scale the ridge functions. See details.

...

Additional graphical parameters for the drawn function. See par.

Details

The values of yshift and yscale determine how ridge functions are shifted and scaled for plotting. This can be used to display the functions over data points for instance. If numeric, a vector can be passed with one value for each plotted function. The vector is recycled if necessary. This indicate the desired mean and standard deviation of plotted ridge functions. Note that this is inverse to the parameters in scale (and xcenter,xscale). If TRUE is passed instead, functions are shifted to the intercept and scaled to their corresponding beta coefficients, placing them on the response scale.

See Also

cgaim for the main fitting function and confint.cgaim for confidence interval computation.

Examples

## Simulate some data
n <- 200
x1 <- rnorm(n)
x2 <- rnorm(n)
x3 <- rnorm(n)
x4 <- rnorm(n)
mu <- 4 * exp(8 * x1) / (1 + exp(8 * x1)) + exp(x3)
y <- 5 + mu + rnorm(n)
df1 <- data.frame(y, x1, x2, x3, x4)

## Fit a model
ans <- cgaim(y ~ g(x1, x2, label = "foo") + g(x3, x4, label = "bar"), 
  data = df1)

## Default plot method
plot(ans)

## Select variable
plot(ans, select = 1)

# Same as
plot(ans, select = "foo")

## Add confidence intervals
ci <- confint(ans)
plot(ans, select = 1, ci = ci)

## Change scale and location
# On the response scale
plot(ans, select = 1, ci = ci, yshift = TRUE, yscale = TRUE)

# Arbitrary scale
plot(ans, select = 1, ci = ci, yshift = 1000)

## Change look

# Main line
plot(ans, select = 1, ci = ci, col = 2, lwd = 3)

# Confidence intervals
plot(ans, select = 1, ci = ci, col = 2, lwd = 3,
  ci.args = list(col = adjustcolor(2, .5)))

# Confidence interval type
plot(ans, select = 1, ci = ci, ci.plot = "lines", col = 2, lwd = 3,
  ci.args = list(col = 2, lty = 4))

## Put curves on the same plot (need to shift and scale)
plot(ans, select = 1, col = 2, ylim = c(-2, 3))
plot(ans, select = 2, col = 4, add = TRUE)