Plot a qgcompmulti fit

Produces a base-graphics display of the fitted marginal structural model (MSM) surface from a qgcomp.glm.multi() fit. The default display is a heatmap over the stored fit-time MSM grid. Contour plotting is also supported, along with a slice-based interval display for MSM predictions.

Usage

# S3 method for class 'qgcompmulti'
plot(
  x,
  style = c("heatmap", "contour"),
  scale = "response",
  grid = NULL,
  interval = FALSE,
  slice = NULL,
  level = 0.95,
  xlab = NULL,
  ylab = NULL,
  main = NULL,
  ...
)

Arguments

x

A fitted "qgcompmulti" object.

style

Character string specifying the surface display style. Supported values are "heatmap" and "contour".

scale

Character string specifying the plotted outcome scale. Version 0.5.0 supports only "response" for plotting. Transformed-scale surface plots are intentionally out of scope; use predict.qgcompmulti() with type = "msm_contrast" and contrast_scale = "estimand" for direct ratio-scale regime comparisons.

grid

Optional user-specified MSM grid with columns psi1 and psi2. If omitted, the stored fit-time MSM grid is used. User-supplied values are interpreted on the MSM coding scale.

interval

Logical; if TRUE, produces a slice-based line display with bootstrap intervals instead of a 2D surface plot.

slice

Optional list with elements var and value describing the fixed MSM coordinate for interval plotting. For example, list(var = "psi2", value = 1) fixes psi2 and varies psi1 over the stored MSM-grid support.

level

Confidence level used when interval = TRUE.

xlab

Optional x-axis label.

ylab

Optional y-axis label.

main

Optional plot title.

...

Additional graphical parameters passed to the underlying base graphics call.

Value

Invisibly returns the structured prediction result used to draw the plot.

Details

Plotting in qgcomp.multi is intentionally prediction-driven: the plotting method delegates the scientific computation to predict.qgcompmulti() and only handles rendering. This helps keep the plotted quantities aligned with the documented public prediction interface.

The default plot(fit) call visualizes the fitted MSM surface over the stored fit-time MSM grid. For quantized fits, the heatmap axes correspond to the quantile-index intervention levels.

For stored q = NULL heatmaps, the plotted axis labels are mapped back to the pooled percentile intervention values used at fit time, even when the MSM itself was fit on centered coordinates. This means there are two scales in play:

• the intervention-value scale, which is what users see on the stored heatmap axes for q = NULL; and

• the MSM coding scale, which is what grid and slice$value use for user-supplied prediction or plotting requests.

interval = TRUE switches to a slice-based line display rather than trying to overlay uncertainty on the full two-dimensional surface. This keeps the uncertainty display readable and aligned with the currently supported MSM interval calculations.

Examples

if (FALSE) { # \dontrun{
dat <- sim_mixture_data(
  n = 400,
  pA = 3,
  pB = 3,
  rho_within_A = 0.3,
  rho_within_B = 0.3,
  rho_between = 0.2,
  psi1 = 0.5,
  psi2 = 0.3,
  psi12 = 0.2,
  seed = 123
)

fit <- qgcomp.glm.multi(
  f = Y ~ X1 + X2 + X3 + W1 + W2 + W3 + C,
  data = dat,
  mix1 = c("X1", "X2", "X3"),
  mix2 = c("W1", "W2", "W3"),
  q = 4,
  B = 100,
  seed = 13
)

# Default heatmap of the fitted MSM surface
plot(fit)

# Contour rendering of the same stored MSM surface
plot(fit, style = "contour")

# Slice-based interval display
plot(
  fit,
  interval = TRUE,
  slice = list(var = "psi2", value = 1)
)
} # }