A data mining approach on poi based traffic flow estimation - an industrial case study

Estimating traffic flow became business critical for the German outdoor advertisement market, because each transaction depends on the reach estimation: How many people are reached by a campaign? Fraunhofer IAIS develops many data mining algorithms that answer both questions for different scenarios. In this work, we present a method that uses no frequency counts but points of interesting locations (POI) and the total number of pedestrians. We utilise model trees to compute frequency estimates and show and evaluate this approach on a public building scenario taken from an industrial project.