ggPD_boot: ggPD_boot
In JBjouffray/ggBRT: Explore and Visualise the Results of Boosted Regression Trees

ggPD_boot

R Documentation

ggPD_boot

Description

Function to draw partial dependency plots with fitted function and confidence intervals obtained through bootstrapping (with replacement) from a boosted regression trees obtained with the gbm.step routine in the dismo package.

Usage

ggPD_boot(gbm.object, predictor = NULL, n.plots = length(pred.names),
  list.4.preds = NULL, booted.preds = NULL, nrow = NULL, ncol = NULL,
  col.line = "darkorange", cex.line = 0.5, type.ci = "lines",
  col.ci = "grey80", cex.ci = 0.3, lty.ci = 2, alpha.ci = 0.5,
  smooth = FALSE, col.smooth = "blue", cex.smooth = 0.3, span = 0.3,
  rug = FALSE, rug.pos = "t", common.scale = TRUE, cis = c(0.025,
  0.975), y.label = "Fitted function", x.label = NULL, ...)

Arguments

`gbm.object`	a gbm.step object (object of S3 class gbm)
`predictor`	select a single variable. Can be either a character name (use "") or a number indicating its ranking in terms of relative influence
`n.plots`	number of predictors to display, ranked by decreasing relative influence (default=all)
`list.4.preds`	the expand.grid list over variable range (from plot.gbm.4list)
`booted.preds`	the array from the bootstrap run (from gbm.bootstrap.functions)
`ncol`	number of plots per column
`col.line`	color of the line
`cex.line`	size of the line (default = 0.5)
`type.ci`	display the CIs either as "lines" (default) or "ribbon"
`col.ci`	color of the CIs
`cex.ci`	size of the CIs (relevant for type.ci="lines")
`lty.ci`	line types for the CIs (relevant for type.ci="lines")
`col.smooth`	color of the smoother
`cex.smooth`	size of the smoother line (default = 0.3)
`span`	span of the smoother (default = 0.3)
`rug`	Logical. If TRUE (default FALSE), add rugs to the plot indicating the actual data points
`rug.pos`	position of the rugs (default = "t"). Can be either "t" (top), "l" (left), "b" (bottom), or "r" (right)
`common.scale`	Logical. If TRUE (default), all the plots have the same y limits
`cis`	a vector to specify the confidence intervals (default=c(0.025, 0.975))
`y.label`	title for the y-axis (default = "Fitted function")
`x.label`	title for the x-axis (by default the predictor name and its relative influence)
`nrow:`	number of plots per row
`alpha.ci:`	transparency of the ribbon (relevant for type.ci="ribbon")
`smooth:`	Logical. If TRUE, (default FALSE) add a smoother to the fitted function