Predictive model for the analysis of academic performance and preventing student dropout using machine learning techniques

School dropout is one of the biggest problems in the country of Mexico, there are several factors that cause it, so it is necessary to propose strategies and lines of action to reduce it. This document analyzes a database with the demographic and social characteristics of high school students, which were collected through the application of school questionnaires and reports, in order to detect the factors that cause students to drop out of school, as well as to identify in time the students who need personalized counseling to offer them educational guidance and prevent them from dropping out of school, this analysis was implemented through machine learning techniques by developing a predictive model with the gradient descent algorithm, from the results to check the forecast errors by applying the mean square error metric, to estimate the possible prediction errors of the model, it is expected to have a great social impact by applying these machine learning techniques in educational community achieving that students can strengthen their comprehensive training, in addition to guiding their talents and interests.