A Machine Learning Approach to Explore and Predict Student Motivation Types

Motivation plays a significant role in shaping students' educational outcomes. Understanding the factors that influence student motivation is crucial for enhancing academic performance and designing effective learning environments. This study utilizes Self-Determination Theory to examine various types of motivation, aiming to develop an integrated framework for analyzing and predicting student motivation. The proposed framework employs multi-source data and evaluates three machine learning techniques: Multinomial Regression, Support Vector Machine, and XGBoost. These models are applied to data collected from a UK-based institution, specifically from the Computer Science department. The findings highlight the superior performance of the XGBoost model in identifying learning analytics characteristics that influence each motivation type, achieving precision ranging from 95% to 100%. Additionally, this study explores the underlying philosophy of using LMS data and its features to classify student motivation, supporting the effectiveness of XGBoost in this context. While the results are promising within the context of a single-institution Computer Science setting, further studies are needed to validate the applicability of this methodology across broader educational frameworks.