plot_heatmap: Heatmap of multiple variable selections ordered by importance

In kormama1/seqknockoff: Sequential knockoffs for continuous and categorical variables

Description

Heatmap of multiple variable selections ordered by importance

Usage

plot_heatmap(S, labels, nbcocluster = c(5, 5))

Arguments

S	list of variable selection indices
labels	labels of actual variable names corresponding to selection indices.
nbcocluster	bivariate vector c(number of variable clusters, number of selection clusters). The former number must be specified less than length(labels) and the latter must be less than length(S).

Details

The list of selected variables is converted to a length(labels)-by-length(S) binary heatmap where each entry is either 1 if variable is selected and 0 otherwise. To help visualize most important variables we perform coclustering of both selections and variables (with the blockcluster package).

Value

plot of heatmap

Examples

library(seqknockoff)

set.seed(1)

S <- list(c(sample(1:10,7), sample(11:20,1)),
c(sample(1:10,7), sample(11:20,1)),
c(sample(1:10,7), sample(11:20,1)))

plot_heatmap(S, labels=1:20, nbcocluster=c(5,3))

kormama1/seqknockoff documentation built on April 11, 2021, 7:44 a.m.