Comparison Of optimization Of Algorithm Particle Swarm optimization And Genetic Algorithm With Neural Network Algorithm For Legislative Election Result

Abstract

General Election is one of the characteristics of a democratic country. One of the countries that embrace the democratic system is the state of Indonesia. Elections are a party of democracy in Indonesia to elect representatives of the people who will sit in parliament and provide great opportunities for the people of Indonesia to compete to appoint themselves to become members of the legislature. Research related to the election has been done by researchers is by using decision tree method or by using neural network. each method has its own weaknesses and advantages, but neural network methods can cover the weaknesses of the decision tree. The result of research using neural network method in predicting election result has accurate result value is still less accurate. In this research, we create neural network algorithm model and optimization with particle swarm optimization algorithm to increase attribute weight to all attributes or variables used, select attributes, and feature selection. whereas the Genetic Algorithm for predicting the performance of generalizations based on static properties of networks such as activation function and hidden neurons will be strong enough to find solutions. After testing with neural network algorithm to produce accurate value of 98.50% and AUC value of 0.982, further optimization done with particle swarm optimization obtained an accuracy of 98.85% and AUC value of 0.996. and then done the optimization testing with genetic algorithm obtained an accuracy value of 96.56% and AUC value of 0.925 So that both methods have a difference of accuracy that is equal to 0,35% and difference of AUC value equal to 0,14.