Fourteen UMD students majoring in mathematics and statistics competed in the MinneMUDAC Data Analytics Challenge.
For the challenge, students were given three weeks to clean, process, and manage several large complex sets of data from an insurance provider (Optum United Health) that contained detailed medical/Rx/provider appointment history information on patients from the across the U.S. identified to have Type II diabetes.Â
Complex Variables
The students were asked to characterize the information (demographic, geographic, medical, Rx, type of treatment, clinic visits, hospitalization, etc.) of patients in the database to determine the importance and predictive value of the variables as related to the cost of managing diabetes care. Students were then required to develop a predictive model to predict who of the patients in a hold out data set would have the greatest cost of care.Â
In addition to the challenge of dealing with real data and complex data bases, students had to use scholarly research from journals and information from organizations such as the CDC, American Diabetes Association, etc. to understand the variables in the data and then had to determine how to use them in developing a predictive model.
Furthermore, the statistical, computational and characterization techniques students could and/or should use was not specified. The teams had to use their knowledge of statistics, machine learning, and other theory to determine the most effective statistical/computational approach in processing, managing, characterizing, and analyzing the data.Â
Three Teams
In the graduate division: Marcus Walker, a 2nd year M.S. student in Mathematics, along with undergraduates Myra Garlid, Nicole Randt and Joe Paulsen. This team was required to compete in the graduate division because they had one graduate student member. Despite having to compete against teams with PhD candidates in Statistics, this UMD team finished in the top 10 of 20 graduate teams. A stellar performance for a team of primarily undergraduates! Each team member was awarded $30 for their success in completing the challenge.
In the Undergraduate Experienced Division: Torri Simon, Daniel Peters, Megan Resch, Samantha Kozimor, and Isabel Ma (see photo above) placed in the top three teams in a field of over 20 experienced undergraduate teams. This team was recognized for their overall presentation and their "serendipitous" analysis of the data in relation to the challenge. Each team member was presented with a $200 cash prize.
In the Novice Division: Sarah Volk, Rachel Neisen, Michael McDonald, Rubing Lin, Kiawen Han gave a commanding performance finishing in the top 50% of the novice teams - a great job for a group of rookies! Each team member was awarded $30 for their success in participating.Â
According to the groups' coach, Assistant Professor Tracy Bibelnieks (pictured above on the far right), it was a job well done by all teams. "It was an absolute pleasure to advise such talented students as they worked through the challenge to their final presentations," said Bibelnieks. "All of the students learned a great deal about working with complex large data sets and realized the challenge, creativity and perseverance needed to attack an open-ended data question such as the one for this competition."
Bibelnieks said the students appreciated the chance to work on something beyond the scope of their typical coursework. Also, the competition provided a great opportunity to network with corporate sponsors and the analytic professionals that were there to judge the event.