turn on suggestions

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

Showing results for

- JMP User Community
- :
- Discussions
- :
- Discussions
- :
- Get the a modeled output in a use-able form?

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

Feb 12, 2016 3:03 PM
(1456 views)

I have a data set that comes close to fitting neatly in a tabulate output. All of the X's are categorical. All of the Y's are continuous. There are some combinations of x's, however that have very few data points or no data points, so I get an unreliable or non-existent Y.

To solve this problem, I want to use Fit Model. I can get a very reasonably looking model

Prediction Expression

One point for example:

But I don't know how to output this efficiently. I could look at every point by hand. Or I could take the Parameter Estimates or the Prediction Expression and write a long series of if/then code. But I figure there must be something I'm missing. What I would like is an output like the tabulate I started with but with but filled in with the estimated values. Is there a way to do that in JMP or am I using the wrong tool?

Thanks.

5 REPLIES

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

hi feldybikes,

Perhaps one option for you could be "Output grid table" under the red triangle of the prediction profiler.

alternatively, one of the LSmeans options can be useful. you will find them under the red triangle of the leverage plots.

best,

ron

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

Thanks very much for the response. I think this is what I was looking for. Didn't think to look in the menu of the profilter.

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

I'm not sure if the values in your first 'Tabulate' are predicted or actual values. In any case, though, if you want to estimate or impute missing cells in this tabulation, you could:

- Use Fit Model for the actual values you have.
- Save the prediction formula as a column back to the table.
- Build 'the same' tabulation using 2. for Y.
- Inspect the values in the cells that were previously '.'

You could use JSL to automate this if you wanted to.

But I'm not sure I have understood your question correctly.

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

Always nice to have options on how to solve the problem.

There shouldn't be any difference between saving the predicted value to a column and saving the prediction formula to a column, right? The formula might just be helpful if I were using that prediction with another data set?

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

Yes, that's correct - If you add a new row to the table that has the required 'x values', then the 'y value' computed by the formula will just appear in the appropriate cell automatically.