基于民航团队旅客销售的组合预测方法分析

Analysis of Mixed Prediction Mode Based on Civil Group Passengers' Revenue

利用Matlab分别用回归分析算法、BP神经网络算法、最小二乘支持向量机算法和组合预测算法对民航团队销售数据进行预测和比较分析,期望为民航销售人员提供更加精准的预测信息,以获得更高的航线收益。结果显示神经网络、支持向量机和组合预测3种算法比航空公司常用的回归分析预测精准度有了明显的提高。支持向量机预测精度相对神经网络稍低,却拥有更强的泛化能力。组合预测能避免单一预测方法的误差,更加适合航线销售人员的实际操作。

Forecasting civil group passengers’ revenue data by regression analysis,Back-Propagation Neural Network algorithm,least squares support vector machines（LS-SVM） and mixed prediction mode in Matlab.Comparing the result and providing accurate information for civil sales staff in network revenue.Study reveals that Neural Network Algorithm,SVMS and mixed prediction mode can better predict revenue data than regression analysis.SVMS can less predict data than Neural Network,but can get better result in other situation.Mixed prediction avoid the error of single prediction method.Sales staff can use this approach in real life.