Algoritmos de Aprendizado de Máquina Para a Previsão de Preços de Venda de Flats em João Pessoa PB

Abstract

The Brazilian real estate market has shown continuous grown, and this scenario is especially notable in João Pessoa – PB, where the construction sector has significantly appreciated in recent years. In this context, accurate property pricing becomes essential to ensure fair negotiations that benefit both buyers and sellers. This study proposes the implementation and evaluation of predictive models to estimate the selling prices of flats in the city of João Pessoa, using machine learning algorithms. Based on data collected from the Viva Real website, seven models were applied and compared: Linear Regression, Support Vector Regression, Decision Tree, Random Forest, XGBoost, LightGBM and Neural Network. The results were analyzed using performance metrics to identify the most effective model for predicting flat prices.