Here is a great presentation by @Phil_Kay using the new Model Screening platform in JMP Pro 16.
The situation involves an example from industrial R&D where a company is trying to develop a process for making a high-value chemical product. There are thousands of possible "ingredient" combinations for this process. They chose 150 combinations at random, and hope to use that information to find the best-yielding combination.
The challenge is to achieve the maximum yield possible. To do that, they built models to understand the effect of each variable on yield, and they used this to predict yield for ingredient combinations that weren't tried. They used Model Screening in JMP Pro 16 to build and test the models, and selected the model that performed best on validation data. From there, they were able to narrow their search for the highest yield on combinations that were predicted to outperform over combinations that weren't.
In this video, you'll see how automated workflows for running, comparing, selecting and exploring models saves time, delivers deeper insight, and drive better decisions.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.