The Application Of Combined Forecasting In The Prediction Of China's Civil Aviation Passenger Traffic Volume

The development of China's civil aviation industry is in the historical stage of rapid development of quantity and rapid improvement of quality.The civil aviation sector that continues to deepen supply-side structural reforms will continue for a long time.As an important indicator of the national economy and people's lives in transport enterprises,civil aviation passenger traffic is an important basis for performance evaluation of road transport organizations.From the civil aviation passenger volume,it can reflect the transportation results of the entire civil air transport sector and the society's demand for passenger transport.Therefore,with the increasing attention paid to the civil aviation industry,the forecast of civil aviation passenger traffic is also very necessary and cannot be ignored.The emergence of the combined forecasting model is to combine different types of single forecasting models,gaining expertise,gathering more economic information and forecasting skills,reducing the systematic errors of forecasting,and significantly improving the forecasting results.In order to make air transport fully play its role in the development of the national economy and effectively promote the rapid and efficient development of air transport construction,it is very important to accurately predict the future development of civil air transport traffic.This article introduces the industry background of the civil aviation industry first.It describes the concepts and calculation methods of multiple individual forecasting methods and combined forecasting in detail,and establishes a model for the nation's civil aviation passenger traffic from 1996 to 2016.Firstly,the exponential smoothing method,the ARIMA model and the grey forecasting method are used to establish a single prediction model,and then the results are obtained through different combination methods such as the equal weight method,the linear combination model,and the Bayesian combination model.Finally,the results of the combined model are used to compare the short-term and medium-term prediction results of the six prediction models using the difference of the Theil and the absolute percentage of the average error as the criteria for evaluating the model's accuracy.The comparison results show that the ARIMA model has the best results in the short-term forecast;The linear combination model using the sum of squares of residuals as the weighting method works best in the medium-term forecast.From the empirical results,it is concluded that the combined model can improve the prediction accuracy of the single model to a certain extent,but it is not absolute;based on the comparison of the short-term and mid-term forecast results,the combined model is more suitable for medium-term forecast;the weight of the combined model selection is different.The effects are different;the model evaluation criteria are different,and the conclusions drawn are different;some combination models can only play an optimal effect after the applicable preconditions are satisfied.