topic Re: JSL, delete a column if there's only 1 'group', labeled in another column? in Discussions

JSL, delete a column if there's only 1 'group', labeled in another column?

aliegner1 — Sat, 10 Jun 2023 23:44:22 GMT

Curious about this.

I've got a dt with several thousand columns that I'm scripting some cleansing for further analysis. The data set is process data, where each row is a unique process run from a given machine. Some machines don't report out all the full list of sensors, so I'm stuck with a few hundred columns only coming from a single machine. I'd like to purge these, as I can't do any statistical comparisons with this data.

Any advice?

Re: JSL, delete a column if there's only 1 'group', labeled in another column?

txnelson — Wed, 16 Feb 2022 22:53:52 GMT

The Missing Data Pattern platform

Tables=>Missing Data Pattern

can identify the rows that have specific patterns of missing values. From that, you can select the rows with the specified patterns, and then delete the rows.

Re: JSL, delete a column if there's only 1 'group', labeled in another column?

aliegner1 — Wed, 16 Feb 2022 22:56:44 GMT

hmm..I'm looking at that now. Not really sure how to replicate that in script?

Re: JSL, delete a column if there's only 1 'group', labeled in another column?

txnelson — Thu, 17 Feb 2022 00:48:56 GMT

All of the JMP platforms are scriptable. And, if you run them interactively, they will allow you to have JMP provide you with the script that will generate the output.

For the Missing Data Pattern, the data table created contains a "Source" entry that has the JSL that will run the platform. Here is a sample of the source code from the simple example I ran the platform on

Data Table( "big class" ) << Missing Data Pattern(
	columns( :height, :weight ),
	Output Table( "Missing Data Pattern" )
)