Nonlinear combination forecasting of measles incidence in Shenyang based on General Regression Neural Network

Accurate prediction of the incidence of measles is an important prerequisite for the prevention and treatment of infectious diseases. Propose to use a combined prediction model. First, analyze the seasonal characteristics of measles to establish the seasonal autoregressive integral sliding average model. Second, use the residual correction method to optimize the gray model to obtain better fitting precision. Finally, input the predicted values of the optimized single-item model to the General Regression Neural Network to output the final measles incidence predictive value. The results show that the combined forecasting model is accurate for trend prediction, the surplus sum of squares, rms error, mean absolute error and mean absolute percent error are lower than the single prediction model, so the combined forecasting model’s prediction effect is good.