DETEKSI PENYALAHGUNAAN NARKOBA DENGAN METODE TWIN BOUNDED SVM

Tomasouw, Berny Pebo
Lesnussa, Yopi Andry

BAREKENG: Jurnal Ilmu Matematika dan Terapan
Vol 15 No 4 (2021): BAREKENG: Jurnal Ilmu Matematika dan Terapan

Twin Bounded SVM (TB-SVM) is an improvement of the Twin SVM method and has advantages in classification problems compared to standard SVM. In this research, linear TB-SVM and nonlinear TB-SVM methods will be applied to detect drug use based on 23 symptoms experienced. The training and testing data is divided into three partition data schemes (60/40 scheme, 70/30 scheme and 80/20 scheme) in order to determine the best level of accuracy that can be obtained. The test results show that the nonlinear TB-SVM with the RBF kernel has a better accuracy rate than the linear TB-SVM, that is 80% at 60/40 scheme, 90% at 70/30 scheme, and 95% at 80/20 scheme.