Comparative Study of Multilayer Perceptron and Recurrent Neural Network in Predicting Population Growth Rate in Brebes Regency

Abstract

Due to its ever-growing population, Brebes had the biggest population in Central Java from 2020 to 2022. The government of Brebes has to predict the growth rate of the population and prepare the resources and employment opportunities to anticipate this population growth rate. This research aims to analyze the result of growth rate prediction in Brebes using Multilayer Perceptron (MLP) and Recurrent Neural Network (RNN). These two methods are applied to determine the most suitable one to predict the population growth rate. This is determined by comparing the smallest MAPE value of these two methods. The analyzed data of the total population from 1991-2022 is taken from Badan Pusat Statistik (BPS) of Brebes. The percentage of division between training and testing data is 80%:20%. According to the research results, the recurrent neural network is the most suitable method, with the smallest MAPE being 1.9973%.

Keywords

• prediction
• population growth rate
• multilayer perceptron
• recurrent neural network