ImpPlot: Plot of Variable Importance

In statVisual: Statistical Visualization Tools

Description

Plot of variable importance based on results from randomForest or gbm.

Usage

ImpPlot(model, 
	theme_classic = TRUE, 
	n.trees = NULL, 
	addThemeFlag = TRUE,
	...)

Arguments

model	An object returned by randomForest or gbm
theme_classic	logical. Use classic background without grids (default: TRUE).
n.trees	integer. The number of trees used to generate the plot used in the function summary.gbm in the R library gbm. Only the first n.trees trees will be used.
addThemeFlag	logical. Indicates if light blue background and white grid should be added to the figure.
...	other input parameters for facet & theme

Value

A list with 9 elements. data, layers, scales, mapping, theme, coordinates, facet plot_env, and labels.

Author(s)

Wenfei Zhang <[e-mail]>, Weiliang Qiu <[e-mail]>, Xuan Lin <[e-mail]>, Donghui Zhang <[e-mail]>

Examples

library(dplyr)
library(randomForest)
library(tibble)


data(esSim)
print(esSim)

# expression data
dat = exprs(esSim)
print(dim(dat))
print(dat[1:2,])

# phenotype data
pDat = pData(esSim)
print(dim(pDat))
print(pDat[1:2,])

# feature data
fDat = fData(esSim)
print(dim(fDat))
print(fDat[1:2,])

# choose the first 6 probes (3 OE probes, 2 UE probes, and 1 NE probe)
pDat$probe1 = dat[1,]
pDat$probe2 = dat[2,]
pDat$probe3 = dat[3,]
pDat$probe4 = dat[4,]
pDat$probe5 = dat[5,]
pDat$probe6 = dat[6,]

print(pDat[1:2, ])

# check histograms of probe 1 expression in cases and controls
print(table(pDat$grp, useNA = "ifany"))

pDat$grp = factor(pDat$grp)


rf_m = randomForest(
  x = pDat[, c(3:8)], 
  y = pDat$grp, 
  importance = TRUE, proximity = TRUE
)


statVisual(type = 'ImpPlot', model = rf_m)

ImpPlot(model = rf_m)

statVisual documentation built on Feb. 21, 2020, 1:08 a.m.