plot_proj and plot_importance are convenience functions providing diagnostic plots for projections, plot_proj plots predictions. This function is tailored for use on ranger projections, with out-of bag predictions for the training set shown in blue, and predictions for points outside the training set shown in black. Either parm or proj will then be needed to identify the projection to be plotted (perhaps both in some programming contexts, or if other projections methods are used).
plot_importance plots the importance metric variable.importance stored in a ranger object. Here, the projection may be identified by either parm or proj, or even directly provided as the object.
plot_proj(
object, parm=NULL, proj,
new_rawdata,
use_oob=Infusion.getOption("use_oob"), is_trainset=FALSE,
xlab=NULL, ylab=NULL, ...) plot_importance(object, parm, proj, n.var = 30L, xlim=NULL,
xlab = "Variable Importance", ylab = "", main="", ...)
These functions are mainly used for their side effect (the plot). plot_importance returns the vector of importance values invisibly. plot_proj returns invisibly, for each parameter a list with the x,y
xlab and ylab elements of the plot call, or a structured list of such lists when plots are produced for several parameters.
An object of class SLik_j. For plot_importance, it may also be an object of class ranger.
Character string: a parameter name. Either one of parm or proj is required to get a diagnostic plot for a specific projector. Otherwise, a multipanel plot will be produced for all projectors.
Character string: name of projected statistic.
Reference table on which projections should be computed. If NULL,
projections are not recomputed. Instead, the values stored in the object are used.
Passed to project.default. Ignored if new_rawdata is NULL.
Integer: (maximum) number of predictor variables to be included in the plot.
Passed to plot (for plot_proj) or to dotchart (for plot_importance).
Projectors may be updated in an object after the projected statistics have been computed and included in the reference table, without any explicit action of the user on the object (see general information about Infusion for why and when this may occur). In particular the projectors stored in input and output fit objects of a refine call are stored in the same environment, and therefore the projectors stored in the input object are modified in light of new simulations. In that case, the plots produced by plot_proj may reflect properties either of the updated projections (if new_rawdata is used), or worse, may mix results of different projections (if new_rawdata is not used). A message or a warning may be issued when such events occur.
## see Note for links to examples.
Run the code above in your browser using DataLab