Share your ideas for the JMP Scripting Unsession at Discovery Summit by September 17th. We hope to see you there!
Choose Language Hide Translation Bar
Highlighted
Camillus
Level II

Standard Least Square (Factor Profiling)

Greetings co-JMP users!

Can you help me regarding with this SLS Factor profiling? JMP predicting over 100% Yield. How can I managed to predict at maximum 100% yield only?

ICHINOSE
2 REPLIES 2
Highlighted

Re: Standard Least Square (Factor Profiling)

It is the result of using a normal distribution for the errors and a technique (regression) that assumes an unbounded response (Y). You have some choices.

 

Live with it! You know that a prediction above 100% is impossible but consider what effect that has on your work. If you maximize the yield and the prediction happens to exceed 100, then you are still likely to have found the real optimum.

 

Clamp it! Save the model as a prediction formula and then enclose it in another function such as Minimum( 100, prediction ).

 

Change it! Use a different distribution model for the errors. The normal distribution is symmetric and extends from negative infinity to positive infinity. Other models are bounded.

 

Censor it! Use a method that respects censoring. (Actually, I am not sure that this approach would help but I am throwing it out there.)

Learn it once, use it forever!
Highlighted
Camillus
Level II

Re: Standard Least Square (Factor Profiling)

Thank you very much! IT really helps me a lot.

ICHINOSE
Article Labels

    There are no labels assigned to this post.