Exploration of a System to Determine the Academic Performance of Engineering Students Through Machine Learning

Abstract

Academic performance is one of many factors used to measure the efficiency of the educational process. It is hereby proposed to improve it by using a prototype that includes data analytics with different actors of the process (student, teacher and institution). The exercise is performed with Industrial Engineering students from Universidad Distrital (Colombia). Although there have been studies on the matter from both quantitative and qualitative approaches, there is still room for research using automatic learning tools. In this work, Python-based tools are used through the specific application of classification algorithms. The prototype design involved diagrams derived from the Unified Modeling Language (UML) and the open use programming language Python. This prototype collects the information from Universidad Distrital which is analyzed using task classification algorithms (decision tree, k-nearest neighbours - KNN, support vector machines – SVM, linear regression, Naïve Bayes, among others) allowing to make initial predictions that ultimately facilitate the decision-making process.