Application of Process Neural Network on Consumer Price Index Prediction

In this paper we present a prediction method of consumer price index (CPI) based on process neural network (PNN). In order to reduce errors, after the raw data was directly expressed as a set of orthogonal basis expanded form, we made use of time-varying input function feature of process neural network and trained process neural network with combined type improved BP algorithm. We achieved a multi-variable CPI prediction with non-linear model of process neural networks gotten by above-mentioned result and illustrated the advantage of process neural network compared to traditional neural network in economic time series prediction. We provide a new method for economic time series prediction in this paper.