Multiple linear regression model to evaluate the market value of residential apartments in Fortaleza, CE

Abstract The valuation of real estate, which assists in the definition of market value, is an important science with a wide field of action, which includes the collection of taxes, commercial transactions, insurance and judicial expertise. This study presents the construction of a linear regression model to determine the market value (dependent variable) of residential apartments in the city of Fortaleza-CE. The studied database presents 17,493 apartments, divided into 227 plan types in a total of 154 projects launched between the years of 2011 and 2014. The model developed was obtained using Multiple Linear Regression associated with the Ridge Regression technique to solve the existing multicollinearity problem. In the analysis of 30 variables (12 quantitative and 18 dummy type qualitative variables), an equation with 6 variables was reached, which meets the theoretical assumptions for its existence.