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
- :
- Can JMP Developers develop an Add-in to make Scori...

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

Dec 10, 2015 7:29 AM
(914 views)

When a model is developed, we save the predictive formula of the model we developed in a column in the data table. If anyone is interested in using the model (i.e. the predictive formula) to score new data with the same variables fields that were used to develop the model, one would either copy and paste the new data for each variable into the model data table and because the variables and data type are the same, the model (prediction formula) will automatically compute a score for those new individuals. For this instance, maybe your new data may contain fewer or more rows than the model data table. Either way, you will have to do some work by deleting the addition rows of the new data had fewer data than the model data table.

Alternatively, with the new data that needs to be scored available containing similar variables and data types as those used to developed/build the model, you can just copy the prediction formula from the selected model of choice and paste it into a new column in the new data table. JMP Pro automatically, using the prediction formula would score this new set of data. Just make sure that every variable in the prediction formula are also in the new table. Thus, there may be other ways to do that out there, but this is how I and based on other advice, I was able to deal with this issue.

This brings me down to the suggestion, could there be another way to make this process more quickly and unique? I am not suggesting that those approaches listed above aren't quick or unique. When one does analysis in JMP, you are fascinated with just how many additional analysis can be done with just the results from an single output by clicking on the red triangle.

Would it be possible for JMP Developers to work on an Add-in tool that can be used to to score new data sets utilizing models already developed in JMP? Something very similar or advance than the SPSS Score Wizard platform?

Jenkins Macedo

1 ACCEPTED SOLUTION

Accepted Solutions

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

Solution

Hi Jenkins,

We have a new feature called Formula Depot to be introduced in JMP 13 Pro that addresses that problem. It provides a central place to store and manage scoring formulas. From there you can apply them to tables, examine the underlying JSL script, and even generate code in different languages (SAS, SQL, and starting with JMP 13, C, Python, and JavaScript as well) to deploy the model to production.

We'll look forward to your thoughts on JMP 13 when you see it!

Cheers,

Nascif

2 REPLIES

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

Hi Jenkins,

We have a new feature called Formula Depot to be introduced in JMP 13 Pro that addresses that problem. It provides a central place to store and manage scoring formulas. From there you can apply them to tables, examine the underlying JSL script, and even generate code in different languages (SAS, SQL, and starting with JMP 13, C, Python, and JavaScript as well) to deploy the model to production.

We'll look forward to your thoughts on JMP 13 when you see it!

Cheers,

Nascif

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

Jun 22, 2016 11:49 AM
(734 views)

I can't wait to upgrade from 12 to 13 once it is release...umm I am sure the license covers upgrades. I could be wrong.

Jenkins Macedo