Feature selection of RAPD haplotypes for identifying peach palm (Bactris gasipaes) landraces using SVM

Abstract

This present work presents a robust system for the feature reduction, using Deoxyribonucleic Acid (DNA) primer. This system reaches up to 100% classes identification based on Support Vector Machines (SVM). In particular, the biochemical parameterization has 89 Random Amplified polymorphic DNA (RADP) primers of Pejibaye Palm races, and it has been reduced to 10 RADP primers. The development of this application provides economic and computational advantages. When it is reduced the number of primers, this application reduces the economic cost, being a process so much cheaper, up to 11.24% from the initial process. On the other hand, the use of our supervised classification system is faster in order to do a method of origin denomination plant certification, due to reduce the dataset up to 11.24%.