FIR as classifier in the presence of imbalanced data

Abstract

In this paper we are investigating the potentiality of the Fuzzy Inductive Reasoning (FIR) methodology as classifier applied to real world dataset. FIR is a modeling and simulation methodology that is best suited for dealing with regression and time series prediction. It has been shown in previous works that FIR methodology is a powerful tool for the identification and prediction of real systems, especially when poor or non-structural knowledge is available. FIR methodology falls under rule base supervised learning techniques. In this study we are studying the performance of the basic FIR classifier applied to imbalanced classification problems and comparing its results with well-known instance based and rule based approaches.