Beer and statistics, better together!
Join us for a meeting of the Willamette Valley JMP User Group at the Lucky Labrador brewpub in Multnomah Village. Come meet other local JMP users, and learn about predictive modeling from JMP’s Director of R&D, Chris Gotwalt. Pizza, beer, snacks, and soft drinks will be provided.
Thursday, August 2
6-8 PM
Lucky Labrador Public House, Multnomah Village
7675 SW Capitol Hwy, Portland, OR 97219
6 to 6:30: Networking
6:30 to 8:00: Presentation
Please RSVP to jordan.hiller@jmp.com by 7/26, so we can get a headcount for catering.
The Essential Modeling Strategies and Techniques of JMP® Pro Made Easy
Chris Gotwalt, JMP Director of Statistical Research and Development, SAS
JMP Pro is a powerful tool for analyzing data, but can appear a bit intimidating to the new user. In this tutorial we show how validation methods streamline the modeling process and make modeling techniques with fancy names approachable and easy to use. Following a few simple guidelines and best practices leads to a simple procedure for trying a variety of models and then either choosing the best one or reusing the data to combine the models into a new one that is superior to each of the individual ones. We will begin with an overview of how to use Validation columns in JMP Pro in combination with a handful of easy-to-understand diagnostic graphs and metrics that are useful for evaluating and comparing models. Then we will apply these ideas to the set of predictive models in JMP Pro that I have found to be the most useful: Regression Trees, Bootstrap Forest, Boosted Tree, Neural Networks, and Generalized Regression. We give an explanation of each of these models in the simplest, most accessible terms possible, and then give clear guidance on how to use them in JMP Pro, while focusing on only the most useful options. At the end of the tutorial we will also give a brief introduction to the Text Explorer platform with an example that shows how to use it in combination with some of the models covered earlier in the presentation to build text regression models and perform a derived sentiment analysis.
