A network based kernel density estimator applied to Barcelona economic activities

This paper presents a methodology to compute an innovative density indicator of spatial events. The methodology is based on a modified Kernel Density Estimator (KDE) that operates along road networks, and named Network based Kernel Density Estimator (NetKDE). In this research, retail and service economic activities are projected on the road network whose edges are weighted by a set of centrality values calculated with a Multiple Centrality Assessment (MCA). First, this paper calculate a density indicator for the point pattern analysis on human activities in a network constrained environment. Then, this indicator is modified to evaluate network performance in term of centrality. The methodology is applied to the city of Barcelona to explore the potential of the approach on more than 11,000 network edges and 166,000 economic activities.