Demo Topic
Predictive Modeling
Purpose
The purpose of this demo is to reveal trade-offs when building predictive models in standard JMP and how JMP Pro can be a more effective tool.
Video
There are two videos. The first is an explanation of the demo for SEs. The second is the demo itself.
Analytical Workflow(s)
Predictive Modeling and Machine Learning
Industry(s)
General purpose.
Questions Being Answered
The target audience are people who have some familiarity with predictive modeling who may benefit from having more tools at their disposal. It is also expected that some people in the audience may not understand what predictive modeling is.
In terms of business case, if a customer is comfortable using fit model or partition, they really should have JMP Pro in their quiver to deal with modeling problems that are not well suited to ordinary least square and basic classification or regression trees. These problems are commonplace in industry. Solving them effectively can add value beyond the cost of JMP Pro.
What is the Story/Scenario
There are two scenarios used for the demo. The first is a variable screening scenario, the second is process optimization. Both scenarios show how results differ in JMP and JMP Pro and explain why JMP Pro results are more realistic.
JMP Features and Platforms
Partition, Bootstrap Forest, Fit Model Standard Least Squares personality and Fit Model Generalized Regression personality.
Discovery Call Preparation, What to Ask and Possible Pain Points to Look For
Listen for anyone who is actively building statistical models or has a leadership role and a vision for using data to solve problems and innovate. Be on the lookout for challenges people have with standard least squares and partition models. For instance, they may have data that are not well suited to standard least squares and have tried transforming response variables without success. Any comments around inaccuracies in model predictions are an opening for a JMP Pro discussion. Bottom line, if a customer or a prospective customer is building statistical models they really should have JMP Pro to handle complex data problems. Everyone has them.
Included Material
JMP Project contains all demo materials.
Leave Behind
The JMP Project is shareable.
Tips for Delivering the Demo
Watch the demo explanation for SEs video.
