Study of clustering trechniques in data mining for climate data

Abstract

The acquisition of useful information from meteorological data dumps is difficult due to the increase in the amount of data stored in JPKM. This is because the parameters and amount of meteorological data are increasing from time to time. This large amount of data has made it difficult to analyze the meteorological data for the purpose of forecasting the rainfall. In the process of forecasting the rain distribution, it is unreasonable to use all the meteorological parameters to do the forecasting. Therefore, one of the ways to identify which parameter gives an impact to the accuracy or performance of the rainfall distribution forecasting is to group the meteorological data. The purpose of this study is to study and compare between two grouping techniques, namely partial and hierarchical method to carry out grouping of meteorological data for the purposes of forecasting of rainfall. The results of this study found that partial groupings were more suitable for use in grouping of meteorological data than hierarchical grouping. In addition, the use of meteorological data attributes within different groups provides better forecasting performance than the use of meteorological data attributes within the same group.