Pemodelan Support Vector Machine untuk Klasifikasi Bakteri Patogen dan Non Patogen Berdasarkan Data Sekuens Genom

Abstract

Bacteria are microorganisms that can be divided into two domains, pathogenic (harmful bacteria) and non-pathogenic bacteria (bacteria are harmless). The purpose of this research is making modeling clasifications of pathogenic bacteria and non pathogenic based on the sequence genom and test the effect of kernels and fragment length of the accuration result. The genome sequence obtained from NCBI with long fragments 100 bp, 400 bp, 800 bp, 1000 bp, and then 5000 bp extraction features done using methods K-Mers and methods of Support Vector Machine (SVM) with 3 main kernel, that is Linear, Radial Basic Function (RBF) and a Polynomial as a method of classifier. From this process, 5000 bp fragment length is obtained with RBF is the highest accuracy reached 96.61%