Comparison of real estate appraisal methodologies based on artificial intelligence

The aim of this paper is to present and compare models that estimate the value of a property according to its characteristics. This information can be required by, for example, potential investors in the real estate markett. In this way, they will be able to compare the offer price of the property with the value obtained by the model, before making investment or disinvestment decisions. In this paper we apply supervised (regression tree, random forerst, nearest neighbour, SVM) and unsupervised (clustering) data analysis methods in order to estimate the value of dwellings located in the city of Madrid, Spain. It is found that the housing characteristic that most influences the price is the surface area, and to a lesser extent the number of rooms, whether the building has a lift, the number of bathrooms and whether the building has a car park. The best method of estimation, according to the MAPE, was the random forest method.