Forecasting inflation and price index with neural networks

The aim of this paper is to forecast inflation and price index in Iran. This data includes annual inflation and monthly data of consumer price index in Iran from 1340 to 1392. In this research، artificial neural network is used to forecast inflation. In this paper are used Error-Back Propagation (BP) with 15 neurons and a radial basis network (RBF) with 10 neurons in the middle layer. For predicting the monthly consumer price index were used the back-propagation with five neurons and neural network with 15 neurons in the middle layer. In the first layers were used sigmoid transfer function and the second layers were used linear transfer function. The remaining 70% of observations is used for testing and training the network. Network input and output price index has been used، consumer price index، with a 12 break in the current period to predict consumer price index in period 12. Based on these results of this paper، it is expected that the inflation reduce to 27. 3 with BP approach.