Classification in point patterns on linear networks under clutter

The problem of features detection under present of clutter in point process on linear networks establishes a methodological and computational challenge with multiple kind of applications as traffic accidents among other. Previous works related to the same topical but developed in more simpler geometries tackles the issue of the clutter removal through the distance of nearest-neighbour and show good results with high classification rates. We extend this procedure to the linear networks motivated by the classification of the traffic accidents on the road network of a city. Simulations demonstrate the performance of the method.