Fork me on GitHub

Visualizing Missing Data

November 17, 2012
Categorized as: R, R-Bloggers

There are several graphics available for visualizing missing data including the VIM package. However, I wanted a plot specifically for looking at the nature of missingness across variables and a clustering variable of interest to support data preparation in multilevel propensity score models (see the multilevelPSA package). The following examples uses data from the Programme of International Student Assessment (PISA; see pisa package).

The required packages can be downloaded from github. Note that the pisa package is approximately 75mb.

> require(devtools)
> install_github('multilevelPSA', 'jbryer')
> install_github('pisa', 'jbryer')

The following will setup the data to be plotted. There is a pisa.setup.R script included in the multilevelPSA package that is included to assist with a demo there. Among many things, it creates a vector psa.cols that defines the variables of interest in performing a propensity score analysis. These are the variables where missingness needs to be addressed.

> require(multilevelPSA)
> require(pisa)
> data(pisa.student)
> pkgdir = system.file(package='multilevelPSA')
> source(paste(pkgdir, '/pisa/pisa.setup.R', sep=''))
> student = pisa.student[,psa.cols]
> student$CNT = as.character(student$CNT)

And finally, to create the graphic use the plot.missing command.

> plot.missing(student[,c(4:48)], student$CNT)

Missing Plot for 2009 PISA

comments powered by Disqus

= Github page; = RSS XML Feed; = External website; = Portable Document File (PDF)
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License. Creative Commons License
Formulas powered by MathJax