摘要
航班预测是航空公司收益管理的关键技术 .本文提出了一种基于C 均值聚类的航班预测模型 ,并将该模型和广泛应用的增量法、回归法进行了对比 .该模型基于聚类方法分析航班销售特征 ,依靠归类决定预测结果 ,屏蔽了日期和季节特性对预测过程的影响 ,降低了算法复杂度 .该模型具有运算速度快、鲁棒性强、预测精度相对较高等优点 。
Flight demand forecasting is the core technology for airline revenue management. This paper presents a new flight forecasting model that is based on C-mean clustering algorithm. The date and season attributes about the flight are discarded, and the complexity is reduced. Compared with the popularly used regression algorithm and pick-up algorithm, the new algorithm is faster, more robust and more accurate, and it has been applied to the real system of Xiamen Airlines.
出处
《信息与控制》
CSCD
北大核心
2003年第6期553-555,560,共4页
Information and Control
基金
国家自然科学基金资助项目 ( 60 1740 2 1)
