A stochastic framework for predicting academic success: Modeling and analysis

This study examines the prediction of academic performance among Yachay Tech University students, focusing on four first-semester subjects. Academic performance, a vital metric in education, is shaped by various factors, making its analysis crucial. To address this, the study employed a combination of methods, including mixture models and the K-Prototypes algorithm, to account for both categorical and numerical variables. Gaussian mixture models and the K-means algorithm were used to classify students based on their grades, while generalized linear models were applied to predict grades in specific subjects. The findings reveal that the majority of students fall within the “medium” academic performance category, followed by the “high” category, with the “low” category being the smallest. Additionally, the analysis highlighted the significant role of categorical variables in predicting grades, with some categories positively influencing outcomes and others showing minimal impact. These results emphasize the importance of categorical variables in shaping academic performance and offer valuable insights for enhancing educational strategies and decision-making processes.