Turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- JMP User Community
- :
- Discussions
- :
- Discussions
- :
- Idea: Contingency table for missing value analysis

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Idea: Contingency table for missing value analysis

Jan 11, 2017 12:33 PM
(1510 views)

I'm looking at the missing values report, and wish that I had a report like the Contingency analysis "contingency table". I would like to know the count of variable "a" that are missing as a function of the other values.

Im looking for something that would allow me to make a working model of the working value clustering so that I can know the probability of one variable having missing values as a function of other variables also having missing values. The clustering picture is nice, but I'm finding it hard to take apart.

2 REPLIES 2

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: Idea: Contingency table for missing value analysis

Jan 11, 2017 12:44 PM
(1506 views)
| Posted in reply to message from engrstudent 01/11/2017 03:33 PM

You could create an indicator data column with a formula for two levels, missing and non-missing. Then you coulld use it in the Y role with the Cotingency platform.

Learn it once, use it forever!

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: Idea: Contingency table for missing value analysis

I stacked my matrix, made a new column with a conditional formula "ismissing(data)" that was 1 for a missing and 0 for non-missing. I then re-assembled the array by splitting, made a multivariate plot and am looking at correlation and partial correlation tables.

It is a hack, but it gets me moving. I have some row indicators that I did not stack, so I was able to re-assemble it.