plot.projsel: Plot of relative explanatory power of predictors

In hsstan: Hierarchical Shrinkage Stan Models for Biomarker Selection

Plot of relative explanatory power of predictors

Description

The function plots the relative explanatory power of each predictor in order of selection. The relative explanatory power of predictors is computed according to the KL divergence from the full model to each submodel, scaled in such a way that the baseline model (either the null model or the model containing only unpenalized covariates) is at 0, while the full model is at 1.

Usage

## S3 method for class 'projsel'
plot(
  x,
  title = NULL,
  max.points = NULL,
  max.labels = NULL,
  from.covariates = TRUE,
  font.size = 12,
  hadj = 0.05,
  vadj = 0,
  ...
)

Arguments

x	A data frame created by projsel().
title	Title of the plot. If NULL, no title is displayed.
max.points	Maximum number of predictors to be plotted. If NULL (default) or 0, all points are plotted.
max.labels	Maximum number of predictors to be labelled. If NULL (default), all predictor labels present in x are displayed, which may result in overprinting.
from.covariates	Whether the plotting should start from the unpenalized covariates (TRUE by default). If set to FALSE, the plot includes a point for the null (intercept-only) model.
font.size	Size of the textual elements (labels and axes).
hadj, vadj	Horizontal and vertical adjustment for the labels.
...	Currently ignored.

Value

A ggplot2 object showing the relative incremental contribution of each predictor starting from the initial set of unpenalized covariates.

hsstan documentation built on June 22, 2024, 12:19 p.m.