|
The purpose of using time series analysis of the dynamics of a three respiratory diseases incidence trends and their impact factors, to explore reasonable forecast model for the region to develop respiratory diseases prevention and monitoring measures provide a basis for decision making, as well as other respiratory disease prediction model The research provides the scientific reference. Methods The analysis methods of time series models (ARIMA product model, exponential smoothing, seasonal cycle model), gray prediction GM (1,1) model of three respiratory diseases (tuberculosis, mumps, measles) from 2003 to 2007 analysis of monthly and quarterly incidence study and build predictive models respectively. Finally forecast model parameter estimation, model diagnostics, model evaluation, selecting the optimal prediction model. Results The study found that: from 2003 to 2007, tuberculosis, mumps and measles incidence cyclical fluctuations, and has a trend change; December 2006 to January 2007 in the area occurred mumps outbreak, the high incidence of the month, the level in the same period in other years, and this year the level of other months; 2007 measles pandemic, led to the rapid rise of the incidence of each month of the year, and living in a high level throughout the year . 2. Analysis of the monthly incidence of three respiratory diseases, this study established the ARIMA product model of the monthly incidence of the three diseases, exponential smoothing models, model fitting accuracy and forecast effect is ideal; analysis of three types of respiratory diseases quarter incidence, the study established a the Three Diseases quarter incidence gray forecast GM (1,1) model, the seasonal cycle model, the model fitting accuracy and prediction are better. Monthly incidence of the conclusions of this study establish the three diseases ARIMA product model, exponential smoothing models, model fitting accuracy and predict the effect is more ideal. Comparison, ARIMA product model is superior to the model fitting accuracy, predict the effect on exponential smoothing model, indicating that the the ARIMA product of the model is more suitable for seasonal fluctuations and trends in monthly incidence. The quarterly incidence GM (1,1) model, the seasonal cycle model, the model fitting accuracy and predict the effect is better, but because only in the past five years of data, is still unable to determine whether they are the pros and cons of this remains to be in further research.
|