Pemodelan Prediksi Hasil Pilkada Dengan Mengunakan Jaringan Saraf Tiruan Back Propagation

Abstract

Now days, on dynamic political environment, PILKADA is one of the hottest issue besides the presidential election. The proposed application intends to assist the stakeholders to react with the dynamic political environment by providing PILKADA prediction. The application is developed based on Artifcial Neural Network framework. Valid data from Komisi Pemilihan Umum (KPU) dan Lembaga Skala Survei Indonesia (SSI) is used as the training and testing data. Lavernberg Marquadt is used as the learning algorithm to tune the ANN model parameters. Based on the simulations, the proposed ANN model based application shows satisfying prediction performance which is indicate by high coefficient correlation (R=0.999). In addition, Lavernberg Marquadt is able to achieve fast convergence which is indicated by low epoch value (epoch=4).