Radial Basis Function (RBF) tuned Kernel Parameter of Agarwood Oil Compound for Quality Classification using Support Vector Machine (SVM)

Abstract

The quality grading of agarwood oil is vital issue among producers. This paper presents the implementation of Radial Basis Function (RBF) tuned parameter in Support Vector Machine (SVM) for agarwood oil quality classification. The work involved of GC-MS based data of agarwood oil, were fed into SVM programming as input and the quality of oil as output. The high and low qualities of agarwood oil were pre-processed using MATLAB software version 2015a which involves of normalization, randomization and data division into training datasets (80%) and testing datasets (20%). By using ‘svmclassify’ script function in MATLAB version R2015a, the data is trained and tested as well as their performances were measured. Several criteria were chosen; specification, precision, accuracy, sensitivity, error rates, error test and mean square error in grading the agarwood oil. It can be concluded that the SVM modelwith RBF tuning was a success and passed all the criteria in classifying the agarwood oil qualities. The significant in this research is the reliable of the SVM handle with RBF as kernel parameter and its finding that contributed to the agarwood oil research area especially in grading system.