Prediction of Air Transportation Incidents Based on Combined Model with Optimal Variable Weights
Hongyun Zheng · Zhongguo anquan kexue xuebao · 2013
To acquire data information of air transport incidents sufficiently and improve the accuracy of forecasting,a combined model with optimal variable weights is used.According to the principle that the absolute value of error at sample point in the combined prediction is minimum,gray verhulst model,Brown exponential smoothing model and nonlinear regression model were selected as single forecasting models to construct the combined forecasting model with optimal variable weights of air transport incidents.According to related data on civil aviation in China from2002 to 2011,applying three single models and the combined model,the incident rates per 10 thousand flight hours of civil aviation in China were predicted.The predict results show that for the next two years,the air transportation incidents rates per 10 thousand flight hours will be 0.347 and 0.331 respectively,there will be a downward trend.While gray verhulst model and regression model can not reflect the fluctuation and volatility of the data,and the lag of exponential smoothing model is obvious,the combined forecasting model is able to overcome these shortcomings.The prediction stability and accuracy of the combined forecasting model are higher than those of the single forecasting models.