Gender identification of sitophilus oryzae using discriminant analysis and support vector machine: A comparison study

Abstract

Sitophilus oryzae (rice weevil) known as severe pest to many stored products, including rice. Gender density of S. oryzae is a critical information in predicting the rate of stored grain lost. However, the techniques used in gender identification of S. oryzae is a destructive technique which involved dissection to identify its reproductive organ. It was a tidious work and very time consuming. Thus, this study focused on the use of non destructive technique which only based on numerical information of S. oryzae morphological features to identify its gender. The numerical information was analysed and tested against two model i.e. Discriminant Analysis Model and Support Vector Machine Model. The result show that rostrum length and width were adequate to be used in the Discriminant Analysis Model for gender identification of S. oryzae with 91% correct classification, while Support Vector Machine Model perform poorly in classification with 62% correct classification. Gender identification of S. oryzae using numerical information features were more accurate and liable compared to normal identification which based on the internal reproductive organ.