• Objectives

The purpose of the project is focused on achieving the implementation of educational interventions to reduce both university dropout and low academic performance, with the purpose of improving the quality and equity in higher education.

• Methodology

To meet the objectives, models for the prediction of academic performance and dropout are built, using personal, social, academic and technological factors. This last set of predictors is made up of digital interactions on educational platforms and social media. The identification of the variables that will be included in the models will be carried out in the first phase of execution, based on the updated review of the scientific literature on the early detection of dropout and academic performance in university studies and, from this first result, the variables that most impact these phenomena will be selected through the innovative data mining technique, Random Forest.

• Previous hypotheses

Our starting hypothesis is that the problem of predicting academic performance and early detection of dropout at the university can be addressed by developing new data mining and multicriteria models that incorporate educational data and new data sources (from digital interactions on digital platforms, educational and social media), as well as recent data processing techniques connected to Big Data and Machine Learning. It is to be expected that , due to the Covid19 pandemic, the information collected on educational platforms will now be much more significant, as its use has greatly increased since the start of the lockdowns. Previously, there were groups or subjects that were not virtualized, and those that were had a much lower activity. This is still true, since full attendance has not been recovered. The information obtained at this time, because it is more abundant and of higher quality, will allow better predictions to be made.