A two-attributes-set spatial clustering algorithm for geography data

Cluster analysis has recently become a highly active topic in data mining research. Cluster analysis is a division of data into groups of similar objects in the spatial and large data sets by fewer clusters. However, existing clustering algorithms for geography data had a common problem that they consider only one set of attributes. Actually, we can divide all attributes into two attribute sets to clustering. For example, Weather Bureau would like to know which regions have similar climate phenomenon, where each weather station are described by latitude and longitude attributes, and measurement of temperature, precipitation attributes. We can discover that two different sets of attributes are required, where one set is spatial attribute, and the other one containing temperature and precipitation is called characteristic attribute. Traditional algorithms do not distinguish the two sets of attributes which lead to low quality spatial clustering results. We propose Two-Attributes-Set Spatial Clustering, generating resulting clusters that can be segmented by characteristic attributes and objects in the same cluster are similar in spatial attributes as well.