This layer adds a raster or heat-map representing a fitted multidimensional effect.
l_fitRaster(pTrans = function(.p) 1, noiseup = FALSE, mul = 1, ...)An object of class gamLayer.
a function from (0, 1) to (0, 1) which takes as input a p-value and returns a value,
alpha, which will be passed on to ggplot2::geom_raster, and will determine
the opacity of the heat-map. The p-value quantifies the significance of the smooth
effect at each location (x1, x2). By default pTrans returns 1, but if we
set it to, say, pTrans = function(.p) .p<0.05 then the regions with p-values
higher than 0.05 will disappear. The zto1 function can be used to specify pTrans
in a flexible way.
if TRUE the fitted effect, mu(x1, x2), will be perturbed with random
noise before being plotted. That is, at each location (x1, x2) a random
variable z(x1, x2) ~ N(0, mul * V(x1, x2)) will be added to mu(x1, x2). Here
V(x1, x2) is the estimated variance of mu(x1, x2) and mul
is a scalar multiplier (see next argument). This is useful for understanding in
which areas the smooth is more uncertain, as these areas will appear more noisy.
positive multiplier that scales the variance of the fitted effect. See the noiseup
argument.
graphical arguments to be passed to ggplot2::geom_raster.
See plot.mgcv.smooth.2D, plot.sos.smooth or plotSlice for examples.