An Evaluation Upon Concept and Techniques of Spatial Data Mining |
A growing attention has been paid to spatial data mining and knowledgediscovery (SDMKD). This paper presents the principles of SDMKD, proposes threenew techniques, and gives their applicability and examples. First, themotivation of SDMKD is briefed. Second, the intension and extension of SDMKDconcept are presented. Third, three new techniques are proposed in thissection, i.e. SDMKD-based image classification that integrates spatialinductive learning from GIS database and Bayesian classification, cloud modelthat integrates randomness and fuzziness, data field that radiate the energy ofobserved data to the universe discourse. Fourth, applicability and examples arestudied on three cases. The first is remote sensing classification, the secondis landslide-monitoring data mining, and the third is uncertain reasoning. Spatial data mining algorithms intensely rely on upon the proficientprocessing of neighborhood relations since the neighbors of numerous items mustbe researched in a solitary run of a normal calculation. Along these lines,giving general thoughts behind neighborhood relations and additionally aneffective usage of these notions will permit a tight joining of spatial datamining algorithms with a spatial database management system. This will speed upboth, the improvement and the execution of spatial data mining algorithms. Inthis paper, we characterize neighborhood graphs and ways and a little set ofdatabase primitives for their control. We demonstrate that normal spatial datamining algorithms are overall backed by the proposed fundamental operations.For discovering critical spatial examples, just certain classes of ways"heading endlessly" from a beginning item are significant. This paper highlights recent theoretical and applied research inspatial data mining and knowledge discovery. We first briefly review theliterature on several common spatial data-mining tasks, including spatialclassification and prediction; spatial association rule mining; spatial clusteranalysis; and geovisualization. The articles included in this special issuecontribute to spatial data mining research by developing new techniques forpoint pattern analysis, prediction in space–time data, and analysis of movingobject data, as well as by demonstrating applications of genetic algorithms foroptimization in the context of image classification and spatial interpolation.