Prediction of Poverty Rates with Nonlinear Autoregressive Neural Networks with External Input

Abstract: The indicator of the success of a region's development is determined by the reduction in the poverty rate of its citizens. Based on these problems, a method is needed to determine the level of poverty in Indonesia, both in urban, rural and national areas, one of which is by using the prediction method. In this study, a Levenberg-Marquardt Neural Network model to predict the level of poverty in Bekasi Regency in the future is proposed, namely nonlinear autoregressive neural networks with external input (NARX). The poverty dataset used is sourced from the Central Statistics Agency (BPS) of Bekasi Regency with test data from 2012 to 2020. The test results show an error of 6.9% compared to the real poverty rate.

Keywords: Poverty Rates, Neural Networks, Bekasi, MATLAB, Levenberg-Marquardt.