Extração de regras Fuzzy por meio da aprendizagem de máquina na informática na educação

Abstract

The evaluation of learners’ achievement is an important phase in the learning in online courses. However, an assessment process may emerge in subjective terms (e.g., Good, Bad). Regards the uncertainty, the fuzzy logic is an appropriate approach for such cases because it can model the imprecise knowledge. The paper approach focuses on guide to construct the fuzzy rules by classification techniques (Decision Tree). The present study runs a validation using the FuzzySD approach, which aims to predict the student dropout phenomena in online courses. In this process, the teacher agreed with 70% of the rules generated by the analysis trees and obtained 19 fuzzy rules. Finally, when asked which the most intuitive tree for interpretation was, the specialist highlighted J48 type.