Using Machine Learning Gradient Boosting to model commercial activities

As urban population grow, commercial activities play a fundamental role in providing goods, services and are part of cities fabric. Policy makers and urban planners need new tools to study these areas. This paper describes how machine learning can be applied to predict the density of commercial activities in cities. A supervised machine learning method called Gradient Boosting was applied to a group of spatial data sets. These were used to fit a model, allowing to analyse their combined interactions and predict commercial activities in London at 1 km grid level for 2010 and 2030. The spatial resolution allowed comparing current and future trends. This type of activity is expected to lower closest to the city centre while increasing in further distant areas. Machine Learning has a great potential as a planning tool and the methodology presented in this paper could be further expanded to other cities.