基于奇异谱分析的航空客运需求分析与分解集成预测模型

考虑到航空旅客运输需求影响因素复杂以及航空客运需求序列非线性非平稳等特征,本文提出了一个基于奇异谱分析(SSA)的航空客运需求分析与分解集成预测模型.需求分析阶段,首先使用SSA对航空客运需求序列进行有效分解,接着借助奇异熵理论,将序列重构为长期趋势项、中期市场波动项和短期噪声项;预测阶段,使用排列熵(PE)判断各重构序列复杂度的高低,并依据序列复杂度分别选择粒子群算法(PSO)和布谷鸟算法(CS)双优化的支持向量回归模型(SVR)或单整自回归移动平均模型(ARIMA)进行预测,结果表明,该分解集成预测模型较ARIMA、SVR等基准模型有着更好的预测性能.

An analysis and decomposition ensemble prediction model for air passenger demand based on singular spectrum analysis

Considering the complex influencing factors of air passenger transport and the non-linearity and non-stationary of air passenger demand series, this paper proposes an analysis and decomposition ensemble prediction model for air passenger demand based on singular spectrum analysis (SSA). In the process of demand analysis, the original air passenger demand series is firstly decomposed into several subsequences using SSA, and then the subsequences are reconstructed into three parts based on singular entropy theory: Long-term trend, medium-term market fluctuation and short-term noise. In the process of prediction, the complexity of each reconstructed part is analyzed using permutation entropy (PE), and the support vector regression (SVR) with double optimization by particle swarm optimization (PSO) and cuckoo search algorithm (CS) or autoregressive integrated moving average (ARIMA) is selected to predict according to the sequence complexity respectively. The empirical results show that the decomposition ensemble prediction model has better prediction performance than ARIMA, SVR and other benchmark models.