Open Collections

Efficiently determining aggregrate proximity relationships in spatial data mining

University of British Columbia

This thesis deals with a nearest-neighbour problem. Specifically, we identify proximity relationships between a cluster of points and nearby features (polygons). Since points are often non-uniformly distributed within a cluster, and since the shapes and sizes of clusters and features may vary greatly, aggregate proximity is the level of expressiveness we need. Additionally, we require that our algorithm be scalable and incremental. The main contribution of this thesis is the development of Algorithm CRH which uses encompassing circles, isothetic rectangles, and convex hulls to efficiently determine proximity relationships via successive convex approximations. Pointwise operations are then performed to compute the aggregate proximity statistics, and to rank the features. Our case study shows that an implementation of Algorithm CRH can examine over 50,000 features and their spatial relationships with a given cluster in less than 1 second of CPU time, delivering expressiveness and scalability. We also show that by using memoization techniques, Algorithm CRH provides for extremely efficient incremental processing.

Thesis/Dissertation

application/pdf

2009-01-27

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/3951

Computer Science

University of British Columbia

UBCV

DSpace