Estimation of rustic housing construction costs through Artificial Neural Networks

Abstract

The Public Administration and the private developer seek to have a prior knowledge of the costs of any building. There are numerous methodologies in the literature that allow the real estate valuation of a property, but it has always been done for urban houses in large or medium-sized cities. This research proposes the use of Artificial Intelligence for the study of rustic houses in small cities such as Caceres (Spain). The research proposes a procedure of Artificial Neural Networks (ANN) to achieve on the one hand, to estimate, through an automatic method, the construction cost of rustic houses and, on the other hand, to identify the most determining attributes in its final price and its marginal weight. The designed ANN establishes as more influential variables in the final price of the property the wet spaces (bathrooms and kitchen), the constructed surface and the age in this order, differentiating themselves from the most determining variables in the price of urban houses in large or medium-sized cities that are the constructed surface and its location.