UCLA

Network point patterns are usually analyzed by methods that assume a continuous plane and Euclidean distance. These methods, however, fail to account for the constraint that network spatial phenomena must lie on a network. This paper proposes three statistical methods, called the network (inter-event distance) H-function method, network (nearest-neighbor distance) G-function method, and network (point-to-nearest-event/empty-space distance) F-function method. We do so by extending the existing H-, G-, and F-functions defined on a continuous plane with Euclidean distance, formulating these methods on a linear network with the shortest-path distance.