Multivariate Time Series Forecasting pada Penjualan Barang Retail dengan Recurrent Neural Network

Abstract

Abstrack – Indonesian retail market is growth along with increasing of population and purchasing power. This opportunity needs to be utilized, but on retail business, sometimes there’s a situation in store where it happens Out of Stock or Over Stock. To handle this problem, we can make a forecast or a prediction about the sales on the future. There are many methods on forecasting, but in general they are divide into statistical methods and computational intelligence methods. This research is aimed to forecast retail sales on each day using Recurrent Neural Network (RNN) as part of computational intelligence method. From this research, we can get a result that on retail forecasting sales case, the accuration performance of RNN is better than statistical method.