GDP Economic Forecasting Model Based on Improved RBF Neural Network

Among the existing GDP forecasting methods, time series and regression model are two most commonly used methods. However, traditional macroeconomic models unable to accurately achieve optimal forecasts of highly complex nonlinear dynamic systems due influence multiple confounding factors. In order solve above problems, a economic based on an improved RBF neural network is proposed. First, main methods analyzed. Then, networks problem that technology cannot handle multi-factor nonlinearities well. Second, further improve convergence speed accuracy learning algorithm, Shuffled Frog Leaping Algorithm with global search capability high practicality fused into training. Finally, build model. The performance was tested using approximation Hermit polynomials Iris classification as simulation examples. experimental results show network-based achieves more accurate than other