Case-based reasoning approach in geographical data mining: Experiement and application

The study deems the CBR approach as a kind of problem-oriented spatial data mining method and provides case-based similarity and reasoning algorithms to extract knowledge from geographical data. First, this paper provides problem-oriented method to represent and organize geographical cases. Second, a rough set theory-based approach was employed to quantitatively retrieve these inherent spatial relationships. Third, a general model was then proposed to calculate the spatial similarity among geographic cases considering different spatial characteristics and relationships of geographical cases. The CBR method was then tested by studying a typical geographic phenomenon, Results of the studies show that CBR method has its advantages in quantitatively analyzing spatial data as well as in solving geographical problems.