Proposição de um modelo de regressão linear para avaliação do valor de mercado de apartamentos residenciais

Abstract

The real state market is an important part of the construction business, which has a major role in the national economy. Beyond the contribution to move the economy of the sector more closely related to construction, it also contributes to booming a number of different areas aligned with the business. Rocha Lima Jr (2014) proposes the exploitation of the real state market through positive waves of the economy, without a big variation on real state values. The assessment of property assists the formation of market prices, and it is an important science and it has a broad field of operation, whether with taxation, commercial transaction or insurance and legal expertise, for instance. Because of its relevance, the NBR 1453 was approved in 2001, regulating concepts, methods and procedures for the evaluation of real property valuers. The work now presented a model of multiple linear regression, adopting the technique of Ridge Regression to surpass the problem of a high multicollinearity between the variables used, being able to predict the market price of a residential apartment in the city of Fortaleza-CE, Brazil. To accomplish that, it was used a database with 17.493 apartments launched between 2011 and 2014, with a total of 30 variables after the necessary transformations to the model. From the developed model, it was possible to estate that the representation of apartments’ marquet price in Fortaleza-CE can be made through an equation using its features as the variables of the model. The study of the database made it possible to trace a view of the change of apartments’ features throughout the period of time studied, as well as a better understanding of the most exploited regions of the city by the construction and incorporator companies.