Improving Affect Detections for K–12 Online Math Learning Using Time-Related Variables

• In Event: Emotions and Motivation in Technology-Based Environments

Abstract

This study focuses on improving affect detection in K-12 online math learning by incorporating time-related variables. Using data from the ASSISTments platform, an exploratory analysis is conducted to identify patterns in affective states based on time factors like time of day, day of the week, and seasons. Five sets of dummy-coded time-related features are constructed and integrated into the baseline model to assess their impact on affect predictions. The results indicate that time-related features significantly improve the model's performance, especially for concentration and frustration. The study also examines the model's performance across different urbanicity regions to ensure fairness and generalizability.

Authors

• Xiner Liu, University of Pennsylvania

• Shuhan Ai, University of California - Los Angeles

• Lisette Enumah