Potential Customer using JMP Trial -- Frequent Freezes with Fit Y by X

Question

We (OVR) have a potential customer that is in the midst of his 30 day JMP trial.  This morning he asked the following question:

"...the whole (JMP) program freezing up on me.  Most recently, this morning I changed some columns to nominal, and was running a Fit y by x analysis when one y response, two x factors, and by a nominal column.  Throughout my work the program froze on me three separate times.

Can there be any issues if I have the file imported and someone else opens and starts working on the database?"

I know the question is a bit cryptic at this point.  Has anyone else heard of this?  Are there other specific questions I can ask or tests I can suggest to help get to the root of his problem?

Thanks,
Jerry

JerryFish · Answer

Mike, Martin, and Peter,

Thanks for all of your replies.  The customer says that the "By" variable is just a "Yes/No" column, so (while I haven't seen the actual data file for confirmation) this would not seem to be the issue.  Good to know for future reference though.

At this point I'm offering to hand him off to the good folks in Tech Support, and the issue must revolve more around specific PC issues (RAM, other apps, etc.) than JMP usage.

Thanks for the ideas!

Jerry

Peter_Bartell · Answer

Jerry:

As Martin alluded too...I've seen the same type of behavior wrt to locking/crashing etc. when taking a continuous predictor and changing the data type to nominal WHEN there are an enormous number of values for the nominal variable now. Especially in the Generalized Regression platform penalization regression platforms. One question to ask the customer is "Why are you considering changing the continuous variable to a nominal variable?" I've seen many customers do this 'just because they can'...and for no good reason from a modeling and process understanding point of view. Often times they just don't understand the distinction between a continuous and nominal predictor variable and how in interprets and uses their respective analyses.

martindemel · Answer

I agree with Mike. Regarding the nominal, sometimes it is a continuous nominal variable, which might have many many different entries and therefore categories. 
My guesses here are also #categories in the nominal variable and the 32bit constraint.

May you can get the data from the customer and test it in JMP 64bit or even better, then you can see the data size and number of categories :)