Avaliação de imóveis usando machine learning: estudo comparativo entre regressores

Abstract

This paper investigates the application of machine learning algorithms in real estate price prediction, using the Ames Housing dataset. Real estate price prediction is particularly interesting because it provides valuable information for both investors and buyers, in addition to assisting in urban planning and the evaluation of housing policies. The methodology included extensive data treatment, with categorical attribute encoding, standardization, outlier removal, and feature selection. The research analyzed a variety of algorithms, ranging from traditional methods such as Linear Regression and Random Forest to advanced techniques such as XGBoost and CatBoost. The aim was to evaluate the comparative performance of these algorithms and assess whether their popularity in the literature reflects in practical results. The results showed that modern boosting methods, such as XGBoost and CatBoost, with R2 scores of 93.84% and 93.33%respectively, surpassing 20 studies found in the literature that used the same dataset. This study provides a solid foundation for future research, with the source code available for reproduction and further experimentation.