Discovering Patterns in Missing Values

[#1]

I am working on analyzing patterns of missing values in various well log data in a particular basin. What are some ideas on the best way to analyze these patterns using JMP. Eventually, I would like to classify these missing values as missing at random (MAR), not missing at random (NMAR), and missing completely at random (MCAR) and define which geological conditions and drilling operations contribute to missing values in logging data (EX: washed out zone, change in formation, cased hole, etc). 

[#2 · Accepted solution]

You probably know already that 'Tables > Missing Data Pattern' will generate another table containing two other graphics scripts. It might be worth using the 'Cell Plot' script, possibly in conjunction with a local data filter, to see what you find. Of course the effectiveness of such a visualisation may be influenced by the number of rows and columns in your data.

[#3]

Thank you! I will try this. 

[#4]

Hi, what do you mean by "Cell Plot Script"? 

[#5]

Nevermind. Just figured it out!

[#6]

To add on to @ian_jmp 's advice and counsel is if you intend to model the data there are several platforms which support JMP's 'Informative Missing' capability. You may want to first decide which modeling technique you'd like to use, then read up in the JMP documentation about how that particular platform implements the general category of informative missing.

[#7]

Okay, thanks for the advice!