My most recent research involves using predictive models for early detection of struggling students and applying counterfactual inference to determine alternative course sequences for helpable students.
In the Summer of 2019, I worked with a team of four students (Nate Braukhoff, Austin FitzGerald, Evan Majerus, and Zhiwei Yang) . We created prediction models to detect struggling students. The purpose of this research is to determine if more informed advising can help students have more success in college. First we explored past student data to determine what percent of students struggle during college. Next we determined if struggle is predictable, how much data is required and how accurate predictions are at different times in the future.
My students won second place in the poster competition at the Wisconsin Technology Symposium 2019. The poster is titled, "Applying Predictive Models to Course Curriculums for Early Struggling Students." https://www.wisys.org/news-media/uw-plattevilles-kaitlyn-timmins-wins-wisys-poster-symposium
The poster can be found at:
To view accurately, the poster must be downloaded.
In the summer of 2020, I worked with another team of four students (Chris Kott, Zack Meyer, Lincoln Schroeder, and Nick Tiede). We continued the research from the previous summer. Using the machine-learning models from the previous summer and creating additional models, we determined alternative course sequences students could take and how well students would do in those course sequences.
The students presented at Wisconsin Technology Symposium 2020.
The poster can be found at:
https://docs.google.com/presentation/d/1U9izL2FhMoXs98p_WOt36H50Ix-TkuJmgeHTpMzm2kQ/edit?usp=sharing
To view accurately, the poster must be downloaded.