This work presents the application of data science techniques aimed at the prediction of student dropout patterns whose case study corresponds to structured information in the sectional UPTC-Duitama. In the application of data science, specialized algorithms were applied for the development of prediction models and data analysis is used. Additionally, a data set was structured whose content has been prepared to be trained. The final result of the research presents a predictive model obtained by means of data science techniques and that was validated by several quality metrics that show the quality of the final model obtained.