REPORT: Explainable AI and Hedonic Rent Models

25 / 08 / 22 - 1 minute read

Explainable AI models answer why, when and where rents grow

Machine Learning (ML) excels at most predictive tasks but its complex structure renders it less useful for inference and out-of sample predictions. This article elucidates and enhances the analytical capabilities of ML in real estate through Interpretable ML (IML).

We compare a hedonic ML approach to a set of model-agnostic interpretation methods. IML methods permit a peek into the black box of algorithmic decision making by showing the web of associative relationships between variables in greater resolution.

We confirm that size and age are the most important rent drivers. Building age is shown to exhibit a U-shaped pattern in that both the youngest and oldest buildings attract the highest rents.

IML methods are also able to visualise how the strength and interactions of hedonic characteristics change over time, besides revealing valuable distance decay functions for spatial variables. 

The new knowledge is used to determine the types of assets that perform best at any given stage of the real estate investment cycle.


Dr. Marcelo Cajias

Decision making in hedonic modeling can be made more transparent with machine learning. Based on a sample of 52k apartments in Frankfurt am Main Interpretable Machine Learning (IML) methods are used to examine feature importance, feature effects and spatio-temporal effects.

Marcelo Cajias

Head of Data Intelligence

Marcelo Cajias heads the Data Intelligence section, which is part of the Investment Strategy and Research team at PATRIZIA. In his role he is responsible for the global portfolio of analytical solutions and dashboards that support strategic investment decisions by means of observed and unobserved machine learning forecast models for various asset classes. Marcelo studied business administration at the University of Regensburg in Germany, majoring in statistics, econometrics and real estate economics. He received his doctorate for his thesis on the economic impact of sustainability on listed real estate companies.

His research has been published in various international journals and he has received the RICS Best Paper Award and the German Real Estate Research Prize.

Augsburg, Germany

Dr. Marcelo Cajias

Head of Data Intelligence