Seleksi Atribut Pada Diagnosis Penyakit Liver Menggunakan Decision Tree Dengan Algoritma Genetika

Abstract

Liver disease is caused by inflammation of the liver. WHO shows nearly 1.2 million people in Southeast Asia and Africa per year die from this disease. Therefore, a diagnosis is needed as soon as possible to get further treatment. To make a diagnosis, a classification algorithm is needed which in this study uses the C4.5 algorithm. However, the algorithm is not optimal for forming a decision tree because it requires loading all cases into memory. Therefore, it is necessary to optimize using genetic algorithms to form simpler rules by selecting attributes and trying various possible combinations of attributes until the most optimal combination is obtained. In the evaluation results, the rules generated by optimization are simpler, namely as many as 32 rules when compared to without optimization, which are more complex, which are 145 rules. Then in the evaluation of accuracy, the rules with optimization resulted in a better accuracy of 70,7% when compared to the accuracy of the rules without optimization of 68,9%