Evaluasi Pembangunan Sistem Pakar Penyakit Tanaman Sawit dengan Metode Deep Neural Network (DNN)

Abstract

The limited knowledge of oil palm farmers on oil palm pests and diseases is related to oil palm productivity. Jambi Province is one of the largest oil palm producers on the island of Sumatra. Usually, to find out the types of pests and diseases in oil palm in the field, farmers need knowledge like that of experts about oil palm diseases. However, the limitation of facilities and capabilities becomes an obstacle. This study offers an expert system to analyze oil palm disease using deep learning. This method is deep learning with excellent accuracy. Various recent studies using DNN state that the classification accuracy results are very good. The data used for the expert system using the DNN algorithm comes from oil palm diagnostic data from the Jambi Provincial Plantation Office. After the oil palm disease diagnosis data is trained, the training data model will be stored for the oil palm disease diagnosis testing process. With a total of 11 classes (Leaf Spot Disease, Anthrox Leaf Blight, Leaf Rust Disease, Leaf Canopy Disease, Bud Rot Disease, Root Rot Disease, Fire Caterpillar or Setora Nitens, Red Mites or Oligonychus, Horn Beetle or Orycte rhinoceros, Bunch Borer Fruits and Nematodes Rhadinaphelenchus Cocophilus), with test variables including the number of classes, TP, TN, FP, FN, precision, recall, F1-score, accuracy, and Missclassificaion rate. The highest accuracy value was 0.88, while the lowest value was 0.83 and the average accuracy was 0.86. This shows that the results of expert system diagnosis on oil palm disease data with DNN are quite good.