Exploration and visualisation of missing data about auunconf HOT 7 OPEN

Snap! I have medical data with missing entries too. I'm interested in being able to visual it and explore clusters of missingness as well as other types of data inconsistencies (e.g. end time before start time).
I am hoping to bring a mockup of the kind of datasets that I use at work.

from auunconf.

The mice package (Multivariate Imputation by Chained Equations in R) has some good tools for imputation (MCAR/otherwise).

Also have a look at VIM::aggr for producing a neat plot of missing data.

e.g. http://www.r-bloggers.com/imputing-missing-data-with-r-mice-package/

from auunconf.

Thank Jonno,

Thanks for that, VIM certainly does have some useful plots, what do you think about incorporating them into ggmissing?

from auunconf.

• Summary of missingness (like the norm or MissingDataGUI package)
• Missingness map: more options for setting order of rows and columns that work for large data
• Vignette
• Enable imputed values to position the points

Keep the package simple. Primary purpose is to make ggplot2 graphics that include the missings in the plot.

from auunconf.

I'm not very familiar with ggmissing, but I'd like to know more about it!

BTW, here is a nice example of a scatterplot with margins for missing values http://kbroman.org/d3panels/assets/test/scatterplot/

from auunconf.

7 votes from the AuUnconf... :) Might be worth continuing discussions around this..

from auunconf.

Nick created a channel on the AuUnconf slack account. Anyone interested can join discussions there also.

from auunconf.