Fuzzy If-then Rule Induction with Cumulative Information Estimations Applied to Real-world Data

Real-world data containing instances corresponding to patients with otoneurological diseases were explored with fuzzy IFTHEN rule induction. It was based on transformation of a fuzzy decision tree made with using cumulative information estimations as the locally optimal criterion at its nodes. This method uses linguistic variables that allow us to naturally model various situations appearing in this data. It also gives classification knowledge in a form of IF-THEN rules that are easily readable and understandable by an expert. Our study shows in comparison to multilayer perception neural networks that classification with the induced fuzzy IF-THEN rules is a useful technique for diagnostics of otoneurological diseases.