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
- :
- How do I get prediction/confidence intervals on predictions from a Bootstrap For...

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

Highlighted

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

Mar 27, 2019 9:01 AM
(1417 views)

Hi all,

I need to get confidence/prediction intervals on predictions from a bootstrap forest model in JMP pro.

I would like to extract formulas and values in data table columns (predictions & confidence intervals) and be able to visualize the intervals in the prediction profiler as well. This is possible for the least square models by extracting the ‘Prediction Formula’ and the ‘StdErr Pred Formula’ from the 'Save Columns' under the Fit Model Platform. Can the same thing be done for the Bootstrap Forest predictions in JMP 14.2?

I would like to have an interactive answer.

Thanks in advance!

1 ACCEPTED SOLUTION

Accepted Solutions

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

I don't think there is any simple way to do this.

A bootstrap forest is an average of multiple individual tree models, each built from a different bootstrap sample (with random selection of a subset of variables at each split).

So you could imagine using the predictions of each individual tree as a way to create bootstrap confidence intervals. You might do this by scripting the platform to fit a single tree bootstrap forest and save the prediction formula. And then repeat this process multiple times to have multiple prediction formulae.

You might like to add your requirement as a request on the JMP Wish List.

Phil

1 REPLY 1

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

I don't think there is any simple way to do this.

A bootstrap forest is an average of multiple individual tree models, each built from a different bootstrap sample (with random selection of a subset of variables at each split).

So you could imagine using the predictions of each individual tree as a way to create bootstrap confidence intervals. You might do this by scripting the platform to fit a single tree bootstrap forest and save the prediction formula. And then repeat this process multiple times to have multiple prediction formulae.

You might like to add your requirement as a request on the JMP Wish List.

Phil