Regresar

Comparative study of data mining techniques to reveal patterns of academic performance in secondary education

Abstract:

The data mining techniques allow for unveiling knowledge from large volumes of information, which have recently been explored in information analysis by educational institutions but already with an increasing demand for this sector to support decision-making. In this research, a methodology for comparing data mining techniques is proposed, which is to be applied to the analysis of academic patrons in students of media education. Multiple methods of selecting attributes are applied to reduce the dimensionality and compare three classifiers and multi-classifiers. The experiments are carried out in a dataset of 285 instances and 36 attributes obtained from an educational survey applied to the students of the School of Education of the University of Barcelona 2017-2018. The best results of classification achieved by the multi-splitter Boosted Tree and Bagged Tree with 93.24% accuracy using the data selected using the BestFirst algorithm.