A graph-based approach for detecting spatial cross-outliers from two types of spatial point events

摘要: Abstract Spatial point events are a series of entities with location information (e.g., longitude and latitude) that describe geographical events, such as crime events. The detection outliers from spatial is very helpful in uncovering unusual phenomena. Existing outlier methods mainly focus on single type In practice, it common two (or more) types can co-occur within certain region. this case, the concept cross-outliers defined considers different simultaneously. This study presents an adaptive graph-based approach to fully accurately detect which categorized into target reference points. First, cross K-function utilized determine whether points positively dependent or not. On basis, cross-neighbourhood relationships between constructed by two-level edge length constrained Delaunay triangulation used quantify positive dependency degree each point. By considering distances local differences respect points, multilevel further employed separate cross-outliers. Experiments using both simulated real-life datasets illustrate proposed method form individual collective high accuracy efficiency. Moreover, there no need input any parameters.