In this paper we focus on a critical component of the city: its building stock, which holdsmuch of its socio-economic activities. In our case, the lack of a comprehensive databaseabout their features and its limitation to a surveyed subset lead us to adopt data-driven tech-niques to extend our knowledge to the near-city-scale. Neural networks and random forestsare applied to identify the buildings’ number of floors and construction periods’ dependen-cies on a set of shape features: area, perimeter, and height along with the annual electricityconsumption, relying a surveyed data in the city of Beirut. The predicted results are thencompared with established scaling laws of urban forms, which constitutes a further consis-tency check and validation of our workflow.