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
- :
- DOE for Discrete Data (Categorical Type Response)

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

Nov 2, 2017 5:28 PM
(738 views)

Hey JMPers!

I am struggling to make a discrete data model on JMP 12. I tried to find the answers with many of indices. Now it seems like I am stuck out.

First I thought the choice design is what I want but it probably not. The structure of data is simple. Here is an example data table. I'd like to check A and E (DF of E should be 40000 you know..) ONLY in ANOVA table. If anybody has solutions, plz make me comfy.

1 ACCEPTED SOLUTION

Accepted Solutions

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

Nov 16, 2017 7:11 AM
(572 views)

Solution

The result you are showing is an Analysis of Variance which is typically shown for analysis with continuous ouputs at least for JMP anyway. The way you originally had everything labelled it appeared that you wanted some sort of nominal regression or at least your value labels would have been recognized in JMP as nominal. To get an answer close to what you are showing you need to convert your response to a continuous value. For example use 1 for Accept and 2 for Reject. These values must be continuous for you to get an Analysis of Variance report that is close to what you are showing. You must also use the counts as Frequency when you set up your model. The attached table has a script for the fit that you can run to recreate the model and see the Analysis of Variance report. Another question for you, what version of JMP are you using? The table I have attached is from JMP 13. There are differences that would need to be described for you to run the attached scripts if you are using and older version of JMP.

I ran the model with X as a categorical and a continous factor. For the continous model X had to added as polynomial to degree 3 to DF correct. The answers were exactly the same either way, but both are different than what you show in your results. So, if these results are still insufficient you will need to provide more information as to how you generated your original fit. What software did you use originally?

8 REPLIES

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

Nov 3, 2017 7:07 AM
(695 views)

This data is categorical: your *A1*-*A4* levels are the categorical predictor **X**, your *Accept*/*Reject* levels are the categorical **Y**, and your counts are the frequencies. Here is the JMP data table:

Select **Analyze** > **Fit Y by X**, select **X** and click **X**, select **Y** and click **Y**, select **N** and click **Freq**. Here is the analysis:

You could also use logistic regression: select **Analyze** > **Fit Model**, select **Y** and click **Y**, select **X** and click **Add**, and select **N** and click **Freq**.

The results are the same, but each platform has different options. The contingency table analysis is for a simple case of one X and one Y. The logistic regression analysis allows for more than one X and terms that are transformations of the Xs, such as X1*X2 or X1^2.

Learn it once, use it forever!

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

Nov 5, 2017 3:55 PM
(655 views)

Thank you for the answer but please check out what I actually meant.

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

Nov 5, 2017 3:28 PM
(658 views)

I meant that I put the data above, then I want to see 'this result' on JMP result.

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

Nov 13, 2017 2:55 AM
(575 views)

Can you plese be more specific on A and E. It's not clear to me from the data you provided. Then we might be able to answer the missing piece, as I believe this information is within the log regression report available, just not all shown in the above screenshot.

KrissKdash wrote:

I meant that I put the data above, then I want to see 'this result' on JMP result.

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

Nov 15, 2017 5:22 PM
(549 views)

A is 'level' and E is 'error' and it is not a log regression report.

I just wrote down the result on spread sheet.

If I make a table with 40,000 rows (4 levels, 10,000 repetitions each), then I can see the result below.

However, I would like to make a table as simple as possbile like,,

1 9930

1 70

2 9936

2 64

3 9940

3 60

4 9925

4 75

and like to check the same result beblow.

Can you show me how to do it?

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

Nov 16, 2017 7:11 AM
(573 views)

The result you are showing is an Analysis of Variance which is typically shown for analysis with continuous ouputs at least for JMP anyway. The way you originally had everything labelled it appeared that you wanted some sort of nominal regression or at least your value labels would have been recognized in JMP as nominal. To get an answer close to what you are showing you need to convert your response to a continuous value. For example use 1 for Accept and 2 for Reject. These values must be continuous for you to get an Analysis of Variance report that is close to what you are showing. You must also use the counts as Frequency when you set up your model. The attached table has a script for the fit that you can run to recreate the model and see the Analysis of Variance report. Another question for you, what version of JMP are you using? The table I have attached is from JMP 13. There are differences that would need to be described for you to run the attached scripts if you are using and older version of JMP.

I ran the model with X as a categorical and a continous factor. For the continous model X had to added as polynomial to degree 3 to DF correct. The answers were exactly the same either way, but both are different than what you show in your results. So, if these results are still insufficient you will need to provide more information as to how you generated your original fit. What software did you use originally?

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

Nov 16, 2017 7:14 AM
(514 views)

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

Nov 17, 2017 5:21 AM
(490 views)