Dropping out in higher education is a serious problem that has been

investigated for decades and causes great harm to individuals, edu-
cational institutions, and society as a whole. This article presents

a case study on the application of the survival analysis combined
with the construction of predictive models in the identification
motivating and deterrent elements of dropout in an undergraduate
program in Information Systems at a public institution of higher

education in Brazil. Methods of educational data mining and pro-
babilistic modeling were applied to student data to model students’

expected completion of the course, semester by semester. The re-
sults of the survival analysis are consistent with the literature and

indicate the greatest risk of dropout is in the initial semesters of
the course, while the identification of characteristics of dropout
makes it clear that the subjects of the first two semesters are the
biggest barriers of the course retaining about 50% of the student
population.