Any body got some journals for some of our JMP Days topics?

Chris_Kirchberg

Hi All,

 

I have a couple of topics coming up soon and I was wondering what people have been doing for the following:

1. Using model visualisation and simulation
2. Mining Data and Building Predictive Models.
3. Making the Most of Modeling and Multivariate Analysis
4. Improving Product Reliability.
5. Using Text Explorer to Extend Analysis.
6. Building Dashboards to Access and Share Updated Analyses.

I have some examples fro 2, 3, 5, and 6, but dont want to reinvent my old journals to fit the JMP Day discriptions if someone has some newer examples that really get peoples attention.

 

Thanks,

martindemel

Hi Chris,

 

this might be duplicate to yours, however I used this material  quite often when the topic is modeling, no matter if Mastering JMP or JMP Days. It has a concept of certain situations coming up quite often, starting from

• Getting motivation using Anscombe data (willl change to dino data soon)
• Unknown Problem in the process - Need to find a fast solution to avoid further scrap and costs --> Decision Tree to understand good &bad Settings based on historic data to give guidance to the people in the production with the goal gaining time for more thorough analysis (using tablet data)
• Multicolinearities avoid finding all potential predictors --> In many cases Bootstrap Forest can help to shade a light into your many parameters (using phil's milling data)
• Model provided optimal settings, but a some material has to be changed and the variation is again too large --> Find through simulation optimal settings for both situations in advance of having a problem (using Sam's machine screw data)
• Model provided optimal settings, but still the variation is too large --> find more robust settings using simulation experiments (using Byron's bioreactor data)
• What if your data is functional --> Understand your process way better than using the mean, min, max of parameters measured over time or frequency domain using Functional Data Explorer (using fermentation sample data with the final result only to create interest - will propable change this for JMP 15)

Depending on the interest you can easily leave things out or advance the modeling options for these.

 

My goal is during a JMP day, not to talk about too much of the different modeling methods but rather give the audience a sense for what good modeling is, why they should care for simulation to check robustness and why they should not be afraid of advanced techniques but be glad that they exist and are more or less easy to use in JMP (Pro). If they go out from this session and say: "Yes, these are similar to my problems I face again and again" or "Just using decision tree can help me so much already" or "We have something where this can help" I've achieved all I wanted and can follow up on these who said that and have these issues.

 

For product reliability you  - from a modeling perspective - may use the maschine screw as well the bioreactor example or as of 15 also model driven multivariate control charts from a quality perspective. If it is about reliability itself, I usually only show life distribution and accelerated life test (if at all) and sometimes reliability growth. The rest is a voice over to understand their situation, pain, implication, need-payoff.

 

Hope this helps,

Martin