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

Showing results for

- JMP User Community
- :
- Discussions
- :
- form of expression

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

Sep 11, 2011 11:20 AM
(622 views)

Hello - I am a novice user of jmp and stats in general. I have run a REML analysis and need to get the predicted equation. While I can view the predicted equation I am a little unsure of the form that it is giving me (shown below).

[Missing image]

My question is what is the proper form of the equation here? I assume exp is exponent but after trying various forms of graphing this it does not seem to fit the data as expected.

I am guessing it is -1.3618145835613 + 0.08192115148914*Ta + (something - a different expression for EM and LA but not sure on the form of the Exp????)

A followup question - I have log transformed the data from within the program - does the program give the prediction equation for the log transformed data or the raw data?

I recognize these are basic questions but I would appreciate some guidance here...

Thanks very much.

-Robin

1 REPLY

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

Sep 15, 2011 4:37 PM
(601 views)

The "match" is giving you an "if-then" type expression within the prediction formula to accomodate different categorical levels. It is part of the overall exponential term so for EM species the formula is "e" raised to the power (-1.36 + 0.08*Ambient Temperature -0.017) and for the LA species the prediced valued is "e" raised to the power (-1.36 + 0.08*Ambient Temperature +0.017). In essence the first term is giving you a grand-average, then you have the effect of temperature and then you have the effect of species.

Regarding the log transformation. Did you perform the transformation in the data table or as part of teh modelling process? If you transformed the data in the table, then as far as the model is concerned the log data is the raw data so the predictions are in relation to the log values. If you transformed the response variable as part of the modelling dialog then the prediction "should" be in terms of the untransformed data. I say "should" because I'm saying what I would expect to happen without double-checking it in the software.

Dave

-Dave