Nearest neighbour clutter removal for estimating features in point process on linear networks

We consider the problem of features detection in the presence of clutter in point processes on a linear network. For the purely spatial case, previous studies addressed the issue of nearest-neighbour clutter removal. We extend this classification methodology to a more complex geometric context, where the classical properties of a point process change and data visualization is not intuitive. As a result, the method is suitable for a feature with clutter as two superimposed Poisson processes on the same linear network, without assumptions about the feature shapes. We present simulations and examples of road traffic accidents that resulted in injuries or deaths in two cities of Colombia to illustrate the method.