|
Purpose dynamics of a regional trend of viral hepatitis and its influencing factors using gray theory method to explore reasonable prediction model for the region to develop viral hepatitis prevention and monitoring measures provide a basis for decision making, as well as other infectious diseases prediction model research provide scientific reference. Classic gray theory method GM (1,1) model and the time series model analysis method (ARIMA model, exponential smoothing method) of viral hepatitis (hepatitis A, hepatitis B) from 2000 to 2008, the quarter, the monthly incidence analysis and build predictive models. Finally forecast model parameter estimation, model diagnostics, model evaluation, selecting the optimal prediction model. Results The study found that: from 2000 to 2008, the incidence of hepatitis A quarter cyclical fluctuations hepatitis B incidence reduced year by year since 2000, the trend changes, but the trend did not change significantly, but the same year relatively stable incidence of each season, no obvious cyclical fluctuations. The incidence of hepatitis A quarter seasonal index model, exponential smoothing model GM (1,1); incidence of hepatitis B quarter, established GM (1,1) model, exponential smoothing models. The research results show that the the model fitting precision and prediction ideal. The incidence of hepatitis A month seasonal index model, ARIMA model, exponential smoothing model GM (1,1); based on the incidence of hepatitis B month, established GM (1,1) model, ARIMA model exponential smoothing model. The research results show that the model fitting accuracy and predict effects are good. Conclusion 1. Incidence of viral hepatitis quarter forecast, the results show that the prediction model fitting accuracy is high season in hepatitis A, GM (1,1) the exponential model accuracy than exponential smoothing model excellent; In hepatitis B, GM (1,1) model fitting excellent accuracy compared with the exponential smoothing model. Prediction of the incidence of viral hepatitis months, the results show that the season in hepatitis A, GM (1,1) the exponential model accuracy highest exponential smoothing model fitting accuracy than ARIMA product model excellent; in beta hepatitis, GM (1,1), ARIMA model fitting accuracy than exponential smoothing model excellent. The results show that relative to the ARIMA model, exponential smoothing models, GM (1,1) model with a small sample of data required (4) data can be modeled, the principle is simple, convenient computing advantages. 4 full information model is not necessarily the best model, the results of this study show that in terms of quarterly or monthly incidence of sequence, full fitting accuracy of the information model is not optimal. Therefore, the time series data is not possible, but according to the historical change in trend predictor to select a representative sample, the establishment of a number of different dimensions of the prediction model, the highest model accuracy test selected precision.
|