niceNaPlot: Plot the amount of missing data in a data set

In ryannick28/CustomFunctionsYrotha: Collection of my custom functions

Plot the amount of missing data in a data set

Description

A function to plot the amount of missing data in a data set. The function plots a grid of observations and variables, indicating the missing data by color. To make the plot more easily readable the function orders the observations according to similar NA-patterns using hierarchical clustering. If the data comprises a large amount of unique NA-patterns, the function will first aggregate the data into partitional clusters using kmeans clustering. Afterwards, hierarchical clustering will be applied within and between clusters to determine a good ordering of the observations. The reason for this procedure is to prevent R from crashing when applying hierarchical clustering directly to large data.

Usage

niceNaPlot(x, IDvar = NULL, show_xlab = TRUE, forceUnaggr = FALSE,
nClust = 100, critVal = 10000, verbose = TRUE)

Arguments

x	A data frame or matrix.
IDvar	Optional character string to indicate the name of the observation-ID-variable in x (see examples). The names in IDvar will be used as labels in the generated plot.
show_xlab	A logical value indicating whether the xlabels are shown in the generated plot.
forceUnaggr	A logical value indicating whether a direct, unaggregated clustering of the NA-patterns should be performed even when there is a number of unique NA-patterns larger than critVal.
nClust	How many clusters to find when performing kmeans clustering of the NA-patterns. Only relevant if more unique NA-patterns than critVal.
critVal	Number defining how many unique NA-patterns are allowed before switching to an aggregated version of the NA-pattern ordering.
verbose	Logical indicating if a progress bar for the kmeans clustering should be shown.

Value

If the function call is assigned to a variable, it will store the (ordered) data with the missing values indicated.

Examples

d <-as.matrix(iris)
n1 <- sample(length(d), size=40)
d[n1] <- NA
niceNaPlot(x = d)
### Applied to data frame with IDvar:
d <- as.data.frame(d)
d$ID <- paste0('S', 1:nrow(d))
res <- niceNaPlot(x = d, IDvar = 'ID')
res

ryannick28/CustomFunctionsYrotha documentation built on June 1, 2025, 4:02 p.m.