Aplicação do modelo de regressão linear múltipla para estimação da produção habitacional (MCMV/CVA e MAP) no Brasil

Abstract

Housing production for planning public housing policies and incentives for the private sector imposes, due to its complexity, the need for reliable results. Therefore, it is necessary to adopt procedures based on scientific methods that minimize the portion of subjectivity existing in the value formation. The regression models are commonly used to associate variables, to explain the influence of each variable in such association, and they are also used to predict new and future results. Thus, this work aims to present the practical application of the multiple linear regression model, a statistical technique that allows estimating housing production (response variable) according to the level of some available explanatory variables. After analyzing the dataset, using the R software, the results allow us to conclude the multiple linear regression model is suitable for estimating housing production, taking into account the region (North, Northeast, South, Southeast and Midwest), the typology (my home my life / yellow-green-house and medium-high standard), and time (measured quarterly). However, we suggest making short-term forecasts since the dataset represents a short time (2019-2020), compromising accuracy in the long-term estimate.