Comparison of Bayes Theorem and Dempster Shafer Methods for Detection Pests of Mayas Rice Plants

Artificial intelligence, especially expert systems, has been widely applied in agriculture to detect pests and diseases that attack a plant. The Mayas rice plant is local in the East Kalimantan region, which has experienced a decline in production due to pest attacks. However, identifying Mayas rice pests is challenging because the symptoms are similar between one type of pest and another. This study used two methods, namely the Bayes Theorem and Dempster Shafer, to determine the highest level of accuracy and the appropriate method for diagnosing Mayas rice pests. The research data consisted of 32 symptoms and ten pests obtained from experts and 50 test data from expert diagnosis and both ways. Comparison testing using the confusion matrix. The results show that the accuracy value for the Bayes Theorem method is 74% and the Dempster Shafer method is 90%, based on the calculation of the confusion matrix. The accuracy value shows that the Dempster Shafer method has the highest accuracy value and is an appropriate method for diagnosing possible types of pests in Mayas rice.