Determinação de modelo de aprendizagem de máquina para precificação de imóveis

Abstract

The use of machine learning techniques is indeed opportune to assist in real estate pricing. Among the various types of machine learning tools, it is necessary to determine which one best meets the needs of determining prices for the real estate sector, which holds notable importance in the Brazilian economy. However, subjective and outdated methods are still practiced in this field. This study aims to define the machine learning model, among those studied, that best predicts property prices. After applying a model where the price is defined by variables representing the aspects considered during negotiations, it was concluded that Lasso Regression performed with the lowest errors in projecting price values.