School's Evasion is a sensitive issue in education: year after year, millions of people leave all kinds of  courses before graduating. In Brazil, it is estimated that the dropout rate reaches 24.3% in Undergraduate Education. For Higher Education it ranges from 12% (public colleges) and 25% (private colleges). In e-learning, these numbers may be even higher.

This is very important for many different reasons. Dropouts may be socially stigmatized and also have difficulties getting good jobs. If they leave during Undergarduate Education, and later decide to go back to school, they will have to enter a group where the other students' age is incompatible with theirs. For the educational institution this is also a problem: a dropout may take up a spot in a course for years, making a huge financial loss for the educational institution.

Faced with all this, what can be done to prevent students from dropping out?

It is undeniable that the answer will vary from institution to institution, campus to campus, course to course, and even between different types of education (classroom or distance), the students' gender or financial condition.

What if you could create an optimized predictive model to identify those students who are about to drop out and then develop a solution for the management of this problem? This is precisely the aim of this project.

Assessing student performance and their behavior in the classroom, or using LMS platforms such as BlackBoard or Moodle, we find clues to identify potential dropouts before they occur, giving a chance to the education institution to act beforehand.