An Add-In For Exploring Interactions in Linear and Logistic Regression Models
Hello everyone,
I've been working on moving my R package, rFSA[CRAN link, github link], to JMP for the last few months. You can read more about FSA and rFSA here. Here is an overview of my JMP Add-In:
During the process of building a regression model, scientists are sometimes tasked with assessing the effects of one or more variables of interest. With an additive regression model, effects (e.g., treatment) are assumed to be equal across all possible subgroups (e.g., sex, race, age). Checking all possible interaction effects is either too time consuming or impossible with a desktop computer. My new JMP add-in implements the Feasible Solutions Algorithm (FSA), which is meant to explore subgroup-specific effects in datasets by identifying two- and three-way interactions to add to multivariable linear and logistic regression models without having to check all possible combinations. The add-in allows users to specify their response variable, fixed main effects, modeling type, and then specify which variables they would like to explore for interaction effects. Users then specify how many random starts of the algorithm they would like to complete, and the order of interactions to explore. Once the user hits "OK" the algorithm will seek to find interactions with a feasibly small interaction p-value to add to the model with the main effects specified. Results return in a data table.
My presentation "Exploring Interactions in Regression Models with JMP and the Feasible Solutions Algorithm" will be streamed semi-live during the upcoming JMP Discovery Summit. During this presentation I will discuss the process of moving my R package to a JMP add-in, and an overview/example of how to use the add-in.
Thank you @Mark_Bailey for your assistance on various .jsl related questions.
The add-in and an example dataset is attached to this post. Please let me know if you have any questions or comments.