Performance Analysis Of Entropy Method In Determining Influence Of Self Organizing Map In Classification Process

Abstract

Self Organizing Map (SOM) is the method to grouping topography shape of two dimension as a map that to get easy monitoring the result of grouping distribution. The process of SOM consist of 4 component, there are : inisialitation, competition, team work, and adaptation. From the fourth component, at the first initialitaion process, in initialitaion value quality beginning vector is according to randomly. The concequency from disseminating randomly is to sensitive forward accuration level because of unexacly in choosing quality beginning with the result that get bad enough of accuration to get better of accuration, we can choose one of method are entropy method. Entropy method is using for qualities or to get level of criteria importance based on atribut of dataset. At this research, entropy method is using to get beginning of qualities to algorithm SOM and to computing the accuration level with qualities of randomly scale. After the test with 3 dataset with total of class and the difference attribute then mean level of accuration to SOM method with entrophy is 67.8401%  and with randomly is 51.1878%. The result is proving that the beginning quality with entropy is better with quality method beginning as randomly