How can I limit the analysis from going into negative values?

[#1]

I want to analyze a 4-factor DOE. However, some responses are negative. This makes my model completely meaningless. These are results that cannot be negative. I used two fields to limit the responses. I added a visual representation of these fields, but they did not work. Can I perform an analysis that does not accept negative responses? Can I set such a restriction? Could you please share any suggestions you may have? It may reduce the accuracy value, but what is important to me is that it provides meaningful results.

[#2]

Hi @Furkan 

Are you saying that the values you have measured have produced negative values or that the predictions are producing negative values on a data formed from all positive responses? If you have negative 'real' values being fed into the model, then it it is going to spit out negative values in the predictions

This discussion board will have some good pointers: https://community.jmp.com/t5/Discussions/How-to-restrict-profiler-to-not-predict-negative-values/td-...

Thanks,

Ben

[#3]

Hi @Ben_BarrIngh 

There are no negative results in my experiment results. But when I analyze them, some of the predicted values come out negative. I mean limiting the values predicted by the model.

Because what I predict cannot be negative.

Thank you very much for your answers. I will take a look at this link.

Furkan

[#4]

Hi @Furkan ,

Got it, a good place to start is looking at the Extrapolation Control settings on the Profiler to see if that helps. What you are looking at is a model that isn't fitting well, hence it predicting values that are 'out'.

Thanks,

Ben

[#5]

Hey @Furkan,

You could also look at using the Box-Cox transformation built into the Fit model platform.

Without transformation:

with transformation