Aplicação de redes bayesianas na previsão de crescimento de fluxos de caixa

• Autor(a) principal: Chagas, Ricardo Pedreti
• Data de Publicação: 2008
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional do FGV (FGV Repositório Digital)
• Texto Completo: http://hdl.handle.net/10438/2026
• topic: Valuation; Finanças; Modelos de previsão; Redes bayesianas; Crescimento; Economia; Fluxo de caixa; Teoria bayesiana de decisão estatística; Previsão econômica

Bayesian Networks may be powerful tools for Financial-Economics modeling. When high degree of uncertainty is present, these tools can be used as strongly helpful advisors in the decision making process. Non-linear relations among variables are normally not captured in traditional linear econometric models. Moreover, specially on situation of crisis or rupture, linear relation do no represent anymore a good proxy to real behavior of financial-economical variables. This contributes for increasing the distance between the theoretical forecasting model and the real data. Throughout this work, we show a methodology for gathering and applying data into Bayesian Networks in order to obtain cash flow growing models for some Brazilian companies and the economical sector they play. Later on, we compare the results of such predictions to the traditional econometric models, and finally to the real data observed in such period. As a conclusion of the study, we make an evaluation of the pros and cons of using Bayesian Network for such application.