Abstract:

Models of Machine Learning (ML) class have been finding more and more applications in a growing number of areas in our everyday life. Models allow us to enhance the automation of processes and, consequently, saves time. One area, where ML may find application, is in the appraisal of real estate values. This paper presents a comparison of performance of different ML models in real estate value estimation. Chapter one presents models selected for the comparison, including in particular: Linear Regression (LR), Random Forest Regression (RFR) and Support Vector Regression (SVR). Chapter two contains a description of the experiment and the model quality assessment coefficients considered, namely: R2, MAE, MedAE, MAPE, MSE and RMSE. Chapter three presents’ evaluations of the results obtained from the models, as measured by the coefficients described earlier. The obtained results allow the conclusion that the Linear Regression algorithm, although the simplest of those evaluated, obtains the best estimates of real estate values.