Autoregressive neural network (AR-NN) modeling to predict the inflation rate in West Java Province

The Autoregressive (AR) model describes the situation where the data in the current observation of a time series depends on the previous observation data. AR models have linearity assumptions. However, in reality there is a non-linear tendency in the data so it needs to be combined with a Neural Network (NN) model. NN models can overcome nonlinear problems in data. The purpose of this research is to build an AR-NN model and apply it to the inflation rate data of West Java Province. The result of this study is an AR(2)-NN model generated by summing the AR(2) prediction results with the residual AR(2) prediction results using a NN model that has a network architecture (4-5-1). The results of data processing show that the AR(2)-NN model is able to increase the level of forecast accuracy from a reasonable forecast to an accurate forecast so that the AR(2)-NN model is better used in West Java Province inflation rate data. This is supported by the smaller MAPE values compared to the AR(2) model. The AR-NN model is expected to be a recommendation for predicting inflation rates in the future.