Peningkatan Akurasi Support Vector Machine pada Klasifikasi Pengaduan Masyarakat Menggunakan Algoritma Firefly

Abstract

Public complaint services need to receive responses promptly from the relevant department. Therefore, a text classification system with best accuracy is needed. One of the methods used is the Support Vector Machine (SVM). SVM is a method that has better performance than other machine learning methods but is influenced by parameter selection and the division of training data and testing data. The Hyperparameter C functions to control the optimization between margin and classification errors. The larger the value of the C parameter, the greater the penalty for classification errors. One approach to optimizing parameter selection is using the Firefly Algorithm (FFA). The number of data samples taken from the service was 1,209 samples with 4 category classes, and the dataset was split into 70% training data and 30% testing data. Testing was conducted by applying the optimized parameters and measuring the accuracy level. The C-best = 0.143 with an accuracy rate of 97.52%, up from 94.21%. This proves that selecting the C parameter using FFA can improve the accuracy of SVM.