A hybrid intelligent system applied to financial statement analysis
Roberto Carlos dos Santos Pacheco, Ana Carolina Bacellar Martins, Ricardo Miranda Barcia, Suresh K. Khator · Proceedings of IEEE 5th International Fuzzy Systems · 2002
Although monitoring the financial health of small firms can be decisive to their success, this kind of organization generally depends on expensive financial consultancy to face such problems. The complexity of the solving process has been a barrier to computational solutions. A close study indicates two main strategies that are adopted by a financial consultant: an inductive strategy (to obtain a first clue of the problem) and a deductive strategy (to elucidate its detailed causes). We developed a hybrid intelligent system composed of a neural network (the inductive module) and a fuzzy expert system (the deductive module) to diagnose and indicate alternatives to financial problems. This system models the traditional methodology of financial statement analysis and gives a financial problem diagnostic, indicating possible alternatives to its solution. In this paper, we present the architecture, the connectionist and fuzzy modules of the system, as well as an example illustrating the computational financial analysis.