Recommended figure dimensions for mfsusie_plot()

Returns a width and height (in inches) sized to the number of panels mfsusie_plot() will draw on the supplied fit. The recommendation accounts for the PIP cell, the per-outcome effect cells, and the per-CS sub-cells when facet_cs = "stack" is used with K >= 2 credible sets. Useful for setting fig.width / fig.height in a knitr chunk so each panel has enough vertical space to render legibly.

Usage

mfsusie_plot_dimensions(
  fit,
  m = NULL,
  facet_cs = c("auto", "stack", "overlay"),
  smooth_method = NULL,
  min_pip_height = 3,
  min_cell_height = 3,
  width = 8
)

Arguments

fit

a fit returned by mfsusie() or fsusie().

m

optional integer index. When supplied, recommend dimensions for the focused single-outcome view.

facet_cs

"auto", "stack", or "overlay". Default "auto" matches mfsusie_plot()'s default.

smooth_method

optional smoother name to resolve facet_cs = "auto" against the smoothed credible bands.

min_pip_height

minimum height (in) for the PIP cell.

min_cell_height

minimum height (in) for each effect cell.

width

figure width in inches. Default 8.

Value

a list with components width, height (numeric, inches) and n_cells (integer, total panel count).

Details

Sizing principle: each cell rendered by mfsusie_plot() is a base-graphics panel with a fixed margin overhead (top title + bottom axis labels = ~1.0 inch) plus the actual data plotting area. min_cell_height is the total allotted height per cell — set it to the desired plotting area (~1.6 in) plus the ~1.0 in margin overhead. The defaults (3.0 in per cell) leave ~2.0 in of plot area, which is the visual sweet spot for the effect curve + lfsr-overlay panels emitted by TI / HMM smoothings. PIP and effect cells use the SAME total height so that the underlying layout() weights (uniform) match the returned height. The recommended figure width defaults to width = 8 inches.

Examples

# \donttest{
set.seed(1L)
n <- 100; p <- 20
X <- matrix(rnorm(n * p), n)
Y <- list(matrix(rnorm(n * 32), n), matrix(rnorm(n * 32), n))
fit <- mfsusie(X, Y, L = 4, verbose = FALSE)
fit_s <- mf_post_smooth(fit)
mfsusie_plot_dimensions(fit_s, facet_cs = "stack")
#> $width
#> [1] 8
#> 
#> $height
#> [1] 6
#> 
#> $n_cells
#> [1] 3
#> 
# }