An interactive game to teach predictive analytics

Abstract

This paper introduces a competitive, interactive classroom game designed to teach predictive analytics by simulating the end-to-end process of building machine learning models. The game is conducted in two rounds engaging students in tasks such as variable selection, model building, parameter calibration, and performance evaluation, all within the context of optimizing a fundraising campaign. Played in undergraduate and MBA programs, the game emphasizes hands-on learning, teamwork, and decision-making under time constraints. Analysis of model performance across two rounds showed that 74.3% of student teams improved their results after the debriefing session, demonstrating the game's effectiveness in reinforcing predictive modeling concepts through iterative feedback. Results from an anonymous survey (n = 78) demonstrated high levels of engagement and learning outcomes, with 93.6% of students recommending the game for future use. Key findings include a mean score of 2.10 (on a scale of –3 to +3) for enhancing the learning experience through competition, 1.94 for increasing interest in analytics careers, and 1.90 for encouraging deeper critical thinking compared to traditional exercises. These outcomes highlight the game's effectiveness in bridging theoretical concepts with real-world applications, encouraging critical thinking, and preparing students for analytics-focused careers.