大数据时代下的航班延误组合预测  

Composite prediction of flight delay during the age of big data

作　　者：杨新湦[1] 王倩[1] 刘俊[1] 张宝成[1] 

机构地区：[1]中国民航大学空中交通管理学院,天津300300

出　　处：《中国科技论文》2016年第19期2205-2208,2242,共5页China Sciencepaper

基　　金：国家自然科学基金资助项目(71571182);教育部人文社科青年基金资助项目(14YJC630185)

摘　　要：为更加准确地预测航班的延误率,基于2015年7月国内所有29.89万次航班数据,对大数据进行数据挖掘处理,筛选可利用的数据,然后对数据进行定量化处理,用量化后的数据进行延误预测。首先,对航班数据进行预处理,并通过影响因素分析确定航班延误的主要影响因素;然后,基于最小二乘法进行参数标定,建立组合预测模型预测航班延误率。采用SPSS对组合预测模型进行求解与检验,得出7月份航班的延误率为43.33%,而全国航班的实际延误率为46.79%,验证得出航班延误判断正确率高达81.86%;组合预测模型对8月份的航班的预测结果显示,该模型是可行的。结果表明,组合预测模型能有效预测航班延误情况,能为航空公司、机场提供快速的短时航班延误决策。To accurately predict flight delays, the available data after the data mining process filtered out and the delay was predic- ted after the quantitative analysis, which was based on 298 900 domestic flights in July 2015. The flight data was preproeessed and the main factors causing the delay was determined through the analysis of the influence factors. The least squares method was used for parameter calibration. A combined forecasting model was developed to predict flight delays and SPSS was used to solve and inspect the model. It was concluded that the flight delay rate in July is 43. 33%, and the actual delay rate of the national flight is 46.79~, which eonfirrned that the correct rate of flight delay was as high as 81.86%. The combined forecasting model for the prediction of the August flight showed that the model was feasible. Results showed that the combined forecasting model can effectively predict flight delays, and provide fast and short term flight delays decision-making for the airline and airport.

关 键 词：航空运输 大数据 航班延误率 组合预测模型 航班预测 

分 类 号：V35[航空宇航科学与技术—人机与环境工程]