Modelo Random Forest aplicado a precificação de imóveis à venda em Aracaju, SE

Abstract

The Brazilian real estate market is considered one of the most promising in the world. According to the Brazilian Association of Real Estate Developers (Abrainc), between the years 2012 to 2022, the Brazilian real estate market was considered one of the most promising in the world. According to the Brazilian Association of Real Estate Developers (Abrainc), between the years 2012 to 2022, the average annual appreciation of property prices was 12.2%. This appreciation makes the market attractive not only for people looking to acquire a property to live in, but also for those who want to turn the property into an investment, either by renting or selling. However, before purchasing a property, it is necessary to conduct an analysis of the factors that make up the property’s price, in order to determine if it is being sold at a fair value and if it is located in areas with potential for appreciation. To assist consumers, data from properties for sale in the city of Aracaju – SE, Brazil, was collected using the web scraping technique. These data were later analyzed and used to develop a prediction model based on the Random Forest machine learning algorithm, capable of pricing properties based on their characteristics. After the analyses, it was found that a property in Aracaju has an average price of R$ 561,000.00. The neighborhoods in the South Zone and Expansion Zone have more expensive housing compared to those located in the Central, West, and North Zones. Regarding the prediction model, it was observed that it showed good predictive power for residences up to R$ 1,000,000.00. The most important factor for pricing in the city of study was the size of the construction area.