Fuzzy inductive reasoning for variable selection analysis and modelling of biological systems

Fuzzy Inductive Reasoning (FIR) is a qualitative inductive modeling and simulation methodology for dealing with dynamical systems. It has proven to be a powerful tool for qualitative model identi cation and prediction of future behavior of various kinds of dynamical systems, especially from the soft sciences, such as biology, biomedicine, and ecology. This paper focuses on modeling aspects of the FIR methodology. It is shown that the FIR variable selection analysis is a useful tool not only for FIR but also for other classical quantitative methodologies such as NARMAX (Nonlinear AutoRegressive Moving Average modeling with eXternal inputs). The tool allows to obtain models that interpret a system under study in optimal ways, in the sense that these models are well suited for predicting the future behavior of the system they represent. The FIR variable selection analysis turns out to work well even in those applications where standard statistical variable selection analysis does not provide any useful information. In this paper, the FIR variable selection analysis is applied to a real system stemming from biology, namely, shrimp farming. The main goal is the identi cation of a growth model for occidental white shrimp (Penaeus vannamei) that allows farmers to plan the dates for seeding and harvesting the ponds in order to maximize their pro ts. FIR and NARMAX shrimp growth models are identi ed, and their prediction capabilities are compared to each other.