Redes neurais artificiais aplicadas às avaliações em massa estudo de caso para a cidade de Belo Horizonte / MG

Abstract

The purpose of this paper is to show the importance of the Artificial Neural Networks ANN application on the study of the marketing phenomena, and particularly concerning its application on the real estate market interpretation in the Evaluation Engineering. These models superiority regarding the estimations by the traditional econometrics has already been proved. The construction of real estate pricing models through conventional techniques as they are currently performed faces problems that reduce the estimation precision, particularly due to the ignorance about the functional form that describes the relation between the variables and due to the difficulty concerning the parameters estimation related to the real estate space distribution. Artificial neural networks application in real estate evaluation offers good perspectives and the outcomes so far have shown that ANN are better computational tools then regression linear for the areas where data cannot be represented solely by linear models. As the determination of real estate value is used in a great number of situations, including for taxation, the occasional mistakes may affect expressive parcel of the population. Therefore, this paper concludes for the importance of researches on the alternative techniques use, such as ANN, rather than those currently in use, so as to improve the real estate evaluation process.