Abstract: | The pair correlation function g(r) is an important tool in exploratory data analysis and model choice in point process statistics. In the case of cluster processes, the behaviour of g(r) for small r is particularly interesting. But just these values of g(r) can be estimated with difficulties only. This paper tries to show that kernel estimators yield reliable results. It is useful to work with variable band widths. An example where the points are positions of pines in a forest illustrates the application of the method. |