Today the greatest technological development in action, common to all sectors, is digitalization. Within the real estate sector, which has always been one of the least innovative sectors,digitalization is bringing radical changes both in terms of production and services. Among the innovations underway, there is the implementation of Automated Valuation Models (AVM) through self-learning techniques. Their use can potentially transform the current state of assessment services.

Traditionally, Automated Valuation Models use econometric models. They are based on hedonic price theory, which assigns a marginal price to each characteristic of the asset or its context. Machine learning models, on the contrary, learn directly from the data and take the form that best suits them.

Many operators in the sector, while admitting that they do not know enough about the phenomenon, are sceptical about the use of AVM. The issue is still the domain of scientific research or of few players. There is a lack of knowledge sharing with the whole appraisal sector. The objective of this research is to provide assessors with some empirical evidence that can help them to gain confidence in these models and discover their potential.