Algorithms for Point Processes Analysis

Bernard Picinbono (1)

1 L2S - Laboratoire des signaux et systèmes

A time point process can be defined either by the statistical properties of the time intervals between successive points or by those of the number of points in arbitrary time intervals. There are mathematical expressions to link up these two points of view, but they are in many cases too complicated to be used in practice. In this article, we present an algorithmic procedure to obtain the number of points of a stationary point process recorded in some time intervals by processing the values of the distances between successive points. We present some results concerning the statistical analysis of these numbers of points and when analytical calculations are possible the experimental results obtained with our algorithms are in excellent agreement with those predicted by the theory. Some properties of point processes in which theoretical calculations are almost impossible are also presented.

Keywords: Counting; Lifetime; Point processes; Renewal processes

Domains: Signal and Image processing; Signal and Image Processing

Bernard Picinbono. Algorithms for Point Processes Analysis. Communications in Statistics - Simulation and Computation, 2013, 42 (1), pp.90-106. ⟨10.1080/03610918.2011.633197⟩. ⟨hal-00793385⟩