736518_en-US

liliconcom

Disappearance of confidence intervals on profiler

Good afternoon (from East Coast) JMP staffs and users!

 

I have three questions on using JMP software.

 

1. I built a customized response surface model with triplicates in the central point.

 

When I open the 'Profiler' from 'Fit model', it shows confidence intervals as shown below.

However, when I open this fitted model through 'Profiler' under 'Graph' tab, it does not show confidence intervals.

 

I also do not show CI in 'Actual by Predicted Plot'.

 

 

Do you know what might have caused this and how to fix this issue?

 

2. When you simulate the variables and responses, does it account for confidence intervals of response?

 

3. Is it possible to simulate the variables based on fixed value of response rather than the response based on maximizing the desirability?

 

Any answers or pointing me to relevant post would be appreciated.

 

Thank you!

Best,

 

Bumjun

 

 

 

 

 

Data Exploration and VisualizationDesign of ExperimentsRe: Disappearance of confidence intervals on profiler

Regarding your question #1: In order to show the confidence interval using Graph -> Profiler, you need to use the formula for the confidence interval. When you save the model equation from the Fit Model platform, this will only save the formula for the mean response. The confidence intervals (upper and lower) are separate formulas and each of them needs their own column. I have pasted in a screenshot below, showing how to save the formula for the confidence intervals from the Fit Model platform.

 

Edit: I've changed the screenshot to something more accurate. @Dan_Obermiller explained it correctly. I think you can get what you want using the first two of the highlighted options.

Re: Disappearance of confidence intervals on profiler

Regarding your question #2: I am not a statistician, but (intuitively) I'd be surprised if the confidence interval played a big role in your simulation. Your simulated response depends on how the variability of the input variables is set. If input variables are provided as "fixed", the response doesn't vary because the input variables are not drawn from a distribution. You can see this in the screenshot below. I've set both input variables to "Fixed", which results in no variation of the response (or, in other words, a constant response) plus the JMP Alert notifying that "all factors and responses are constant functions. Therefore, I would answer a loud and clear "No" to your question #2.

 

Re: Disappearance of confidence intervals on profiler

Regarding your question #3: If you have a model for a given response, why (and how?) would you fix it to simulate the variability of the input variables? If you are interested in the variability of your input variables I suggest

• getting experimental data to estimate variability of the input variables
• fit a separate model to making what is an input variable in one model a response variable in another model for estimating variability

Again, not a statistician, but simulating the input variables using a fixed response goes contrary to how I understand model fitting.

Re: Disappearance of confidence intervals on profiler

Thank you Ressel,

 

I saved CI as a formula but still did not figure out how to apply this to the Profiler.

 

When I include those Lower and Upper CI on the Profiler as well as response, it does generate profiler.

Do you have any clue what I am doing wrong?

 

Re: Disappearance of confidence intervals on profiler

For the confidence intervals, save the StdErr Pred Formula. Put the prediction formula for the mean AND the StdErr Pred Formula in the Y area of the Graph > Profiler dialog. JMP will then ask if you want to use the StdErr column to make intervals. Say YES, and you will have your graph.

Re: Disappearance of confidence intervals on profiler

Hi Dan,

 

Thanks for the suggestions.

 

I just tried it as below.

 

However, it does not if I want to use it for CI; it spits out separate profiler for SE as below.

 

 

Re: Disappearance of confidence intervals on profiler

Example to your question #1 attached.

1. Save prediction formula (see screenshot under 2.)
2. Save StdErr Pred Formula
   
   
   
3. Go to Graph -> Profiler, select the columns as shown below into the Y, Prediction Formula role and confirm with OK.
   
   
   
4. The below window will show up. Confirm with Yes.
   
   
   
5. The Profiler created (right) should look exactly like the one in the Fit Model platform (left):
   

Thanks @Dan_Obermiller and apologies for adding to the "noise".

Re: Disappearance of confidence intervals on profiler

In the simulator, below the profiler there is a spot labeled Responses. In there is a weight field where you can enter the noise to add to the response. Change that to be Random by Model and will add your model error to the simulated value. This will give you noise due to the input variables changing + the noise of the response itself.

Re: Disappearance of confidence intervals on profiler

You must make sure it is the PredSE Formula that is saved, not just the values.

Re: Disappearance of confidence intervals on profiler

Thank you guys but it still dose not work.

 

@Dan_Obermiller, I double checked my previous column and it was saved as PredSE Formula.

 

@Ressel When I run Profiler with Pred Formula and PredSE that you saved as a column, it works as you have shown.

 

Strangely, if I make my own PredSE, which is on the last colum on the table below, then I run Profiler with 'New PredSE' rather than the one you created, it runs into the same problem. I am not sure if there is something wrong with some sort of default setting of my JMP software. @Dan_Obermiller  Do you have any insight in it by chance?

 

 

Here I attach the jmp file with new PreSE column I saved from my jmp software.

 

 

 

 

 

 

 

 

 

 

 

Re: Disappearance of confidence intervals on profiler

After looking at your table, I think I see where all of the confusion is coming in!

Using the Graph > Profiler, I put in your Pred Formula Y and PredSE Y columns in the the Y field. JMP then asks if I want to use PredSE Y as intervals. I say yes, and I get this plot:

You can't see the intervals, but if you adjust the Y axis, you can see them.

 

Now if I start over and put in the Pred Formula Y and PredSE Y 2, it does not work. Why? Well, it turns out that there is extra information in these columns that links them together. Open up the Column Info for the Pred Formula Y and click on the Predicting column property. You will see there is an ID number embedded in that property.

If you look at the Column Info for the PredSE Y column, then you will see this same ID number in the Std Error property:

But the PredSE Y 2 column has a different ID number. If you change the ID in the PredSE Y 2 so that it matches the Pred Formula Y, then the confidence intervals will show up when using Pred SE Y 2 (after adjusting the Y-axis).  In other words, JMP is checking to make sure that the potential interval formula actually matches with the prediction formula. This will allow you to plot confidence intervals with multiple prediction formulas simultaneously.

Re: Disappearance of confidence intervals on profiler

Thank you @Dan_Obermiller ,

 

It finally worked after changing the ID on PredSE identical to Pred Formula.

 

Do you consider it as a bug? There will be incidences that people recreate/modify error column.

 

 

Re: Disappearance of confidence intervals on profiler

Maybe. The model error comes from the Root Mean Square Error (RMSE) of the model. So in that model fitting window go to the red triangle for the model and choose Regression Reports > Summary of Fit. Is there a non-zero number in the RMSE field? Also, that menu choice is only available from the Fit Model platform. I don't believe you get that option from the Graph > Profiler option because that is not a model fitting platform.

Re: Disappearance of confidence intervals on profiler

Hmm, interesting. It seems like you'll have to save the Prediction Formula and the StdErr Pred Formula in the same session, i.e. from the same instance of the Fit Model platform.

Click on the star / asterisk next to the column name of the saved columns and select Predicting for the Prediction Formula column. Then do the same for for the PredSE column but select Std Error. (see the screenshots below).

For each of the formula columns you saved, you'll find an ID in parenthesis (see screenshot below). If these IDs don't match up, the Graph Profiler will not combine the prediction formula with the standard error formula. For example, if you first save one of the formula columns from the Fit Model window, then close and re-open the Fit Model window to save the other formula column, the formula columns will have different IDs and the Graph Profiler will not combine them. 

 

However, you can copy paste the IDs between the different columns to make Graph Profiler comply. I have done this in the example below. It is the same as the previous screenshot, but now I have copied the ID from Pred Formula Y 4 to PredSE Y 7 and Graph Profiler gave the desired result.

 

Re: Disappearance of confidence intervals on profiler

Thank you @Dan_Obermiller 

 

Indeed, this option is available only in Fit Model platform.

 

Thank you for your help again!

Re: Disappearance of confidence intervals on profiler

No, this is certainly not a bug. As my last sentence stated, by using this ID, JMP allows you to plot multiple prediction formulas with their corresponding confidence intervals on the same profiler. Think of the situation with Y1 and Y2, with the corresponding StdErr Y1 and StdErr Y2. If I wanted both Y1 and Y2 on the profiler with their confidence intervals, the software could easily get the confidence intervals mixed up. The ID code will prevent that from happening.

 

Further, the prediction formula and the PredSE Formula are typically saved in the same modeling session as @Ressel stated. If you save them in separate sessions (which would lead to different ID numbers), you run the risk of the models not being exactly the same. That means that the confidence intervals would not be correct for the model that you are displaying. The approach that is used by JMP helps to ensure accuracy.

Re: Disappearance of confidence intervals on profiler

@Dan_Obermiller 

 

For accounting for noise of input variables, I see this Random by Model is greyed out.

 

Do you also have troubleshoot for this?