Stochastic modeling and predicting of academic performance

Academic performance is a crucial indicator in education, susceptible to being influenced by various factors such as demographic location, type of school, employment, and family environment, among others. Given that variables impact this indicator, mathematical tools like stochastic models are employed, capable of revealing relevant data to identify relationships among influential factors. Among the prominent methods in this research for detecting different classes are Mixture Models and the K-Prototypes algorithm, specifically for categorical and numerical variables. Gaussian mixtures and the K-means model are employed to classify based on numerical variables. Additionally, Generalized Linear Models are used to predict grades in different subjects. This study focuses on students at Yachay Tech University, providing relevant information for future decisions regarding teaching four subjects in the first semester and helping students understand the factors influencing their academic performance. Programming languages such as Python and R-project were employed to carry out the mentioned models. As the main findings, it is highlighted that the “medium” academic performance category has the highest number of students, followed by the “high” category and, finally, the “low” category. Additionally, when predicting grades in different subjects, it was observed that the different categories of variables had an impact, whether positive or negative, thus indicating the importance of certain categorical variables.

• Glavni autor: Morejón Campoverde, María Mishel (author)
• Format: bachelorThesis
• Jezik: eng
• Izdano: 2024
• Teme: Rendimiento académico; Mezclas Gaussianas; Modelo lineal generalizado; Academic performance; Gaussian mixture; Generalized linear models
• Online pristup: http://repositorio.yachaytech.edu.ec/handle/123456789/729