Iterative local gaussian clustering to extract interesting patterns on spatio-temporal database

The study of spatio-temporal data mining in extracting and analyzing interesting patterns from spatio-temporal database has attract great interest in diverse research field. Huge amount of research has been done in either spatial data mining or temporal data mining and numbers of clustering algorith...

Full description

Saved in:
Bibliographic Details
Main Author: Aman, Tirwani
Format: Thesis
Language:English
Published: 2009
Subjects:
Online Access:http://eprints.utm.my/id/eprint/18367/1/TirwaniBintiAmanMFC2009_TheStudyOfSpatio-TemporalDataMining.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The study of spatio-temporal data mining in extracting and analyzing interesting patterns from spatio-temporal database has attract great interest in diverse research field. Huge amount of research has been done in either spatial data mining or temporal data mining and numbers of clustering algorithms have been proposed. However, not much research has been done in the integration of both spatial and temporal data mining, which is spatio-temporal data mining. The focuses of this study is to analyses the Iterative Local Gaussian Clustering (ILGC) algorithm and implement the algorithm to the spatio-temporal data, which is crime data. . In ILGC approach, the K- nearest neighbor (KNN) density estimation is extended and combined with Gaussian kernel function, where KNN contribute in determining the best local data iteratively for Gaussian kernel density estimation. The local best is defined as the set of neighbors data that maximizes the Gaussian kernel function. ILGC used Bayesian rule in dealing with the problem of selecting best local data. To test and validate the ILGC approach, other clustering method, which is K-Means and Self Organizing Map (SOM) will be implemented on the same data sets.