RELATIONSHIP BETWEEN DATA REDUCTION AND PERFORMANCE IMPROVEMENT OF CLASSICATION WITH K-SUPPORT VECTOR NEAREST NEIGHBOR
K-Support Vector Nearest Neighbor (K-SVNN) as a Nearest Neighbor-based method can be used for data reduction. Reduction is done by giving parameter K as nearest neighbor used. Besides data reduction, K-SVNN can also improve the performance of prediction accuracy. The author tests the value of K that is the percentage of the amount of data. The K used varies from 10% to 100%. This study was conducted to observe the relationship between data reduction and performance improvement. Performance improvement is measured by Fisher Discriminant Ratio (FDR). From the results of research proved that in some data sets, data reduction with K-SVNN can increase the FDR value, while some other data sets can’t. The K value that gives the highest FDR value is 30% to 50% with data reduction up to 4.49%.