Academic performance problems : a predictive data mining-based model

Abstract

Often times, universities are not able to deal with the variety of factors that may affect the academic performance of students. This kind of situation generates the need for tools that establish academic performance patterns, setting profiles as a basis to detect potential cases of underachieving students who need support in their academic activities. This paper proposes the use of Data Warehousing and Data Mining techniques on performance, social, economic, demographic and cultural data from students who took “Algorithms and Data Structures”, which is a subject in the Information Systems Engineering curricula at UTN-FRRe (Resistencia, Chaco, Argentina) in an attempt to establish generic academic performance profiles. From the descriptive analysis obtained during the 2013 to 2015 period from the subject aforementioned, a predictive model was used. It establishes the possibility of students' academic failure, taking into account the factors earlier mentioned.