Abstract:
© 2019 ACM. Learning Analytics (LA) is an analysis toolset that enables collection of students' data and context data, for the purpose of visualizing indicators of performance that allow for improvements for learning and academic success. In this study, artificial intelligence (AI) algorithms like K-Nearest Neighbor and Random Forest were used. These algorithms trained a model that could predict the academic success of college-level engineering students. Under an experimental model with 182 students, three instructors leading six groups of Physics II for engineering majors of the Tecnologico de Monterrey (Mexico) administered adaptive measures for one group each and not for the other of their groups (the control groups). Three forecasts were calculated considering structured academic information (numerical grades) and unstructured academic information (student ID pictures). Unstructured data from the facial photograph and numeric academic information from the first evaluation period were considered for the study. The results show a significant difference between the control and experimental groups of only one instructor, while the results of the other two instructors' control and experimental groups were consistent. One finding of this study is that, despite the prediction not being correct for each student, a general picture of the performance of the group was given. It seems that the algorithm must be trained with more data for the forecast to be more precise in the future.