topic Re: JMP crashing when trying to do Continuous Fit All for 1,152 distributions in Discussions

JMP crashing when trying to do Continuous Fit All for 1,152 distributions

ibrown — Mon, 13 Nov 2023 23:50:38 GMT

I have 288 bins with 4 distributions each (1,152 total) and I'm trying to find the best continuous fit for each distribution. I've tried to press Continuous Fit > Fit All while holding down the Command button (I'm on Mac) so it runs it for all distributions, but this always crashes JMP because that's a ton of calculations. When JMP freezes, I've even let my computer sit for several hours for these calculations to run in the background but no luck.

I know that my distributions will likely mostly be Normal, Normal 2 Mixture, Normal 3 Mixture, and SHASH. I've tried to do one type of fit at a time for all my distributions, but it crashes when I do SHASH. Can anyone help me with a way to do this? Maybe write me some JSL code please? Also, bonus points if anyone can help me export the distribution parameters for all of the best fits into a table.

Note: I'm using JMP Pro 16.

Re: JMP crashing when trying to do Continuous Fit All for 1,152 distributions

Mark_Bailey — Tue, 14 Nov 2023 13:45:58 GMT

I do not understand the context of your data when you say, "288 bins with 4 distributions each."

Do you expect all 1,152 cases to be independent? Do you have any prior knowledge to suggest a distribution model for each or a common model for some of them?

You can right-click on a table of parameter estimates and select Make Into Combined Table. I'll take the bonus points.

You might contact JMP Technical Support (support@jmp.com) about the crashes. They track crashes. It might be a known issue. There might be a workaround, too.

Re: JMP crashing when trying to do Continuous Fit All for 1,152 distributions

Mark_Bailey — Tue, 14 Nov 2023 14:04:15 GMT

I just ran this test. It took about five minutes to complete. It did not crash. I am running the current version, JMP 17.2, on a Windows 10.

Names Default to Here( 1 );

data = J( 100, 1152, Random Normal() );

dt = As Table( data );

names = dt << Get Column Names();
cols = dt << Get Column Reference( names );

dist = dt << Distribution( Y( Eval List( cols ) ) );

dist << Fit All;

What is your data like? How many rows do you have?