偏最小二乘回归在军用飞机价格预测中的应用

由于军用飞机性能要求的不断提高,影响飞机采购价格的费用驱动因子繁多,使得原有的价格预测模型已经不适用于现代军用飞机。分析了军用飞机采购价格样本数据少、费用驱动因子多的特点,考虑到偏最小二乘回归(PLSR)方法在处理小样本多元数据方面的优势,应用PLSR对军用飞机采购价格进行预测。PLSR首先提取第1、第2主成分对采购价格样本的特异点进行剔除;然后进行变量投影重要度分析以筛选费用驱动因子;最后,PLSR对费用驱动因子进行回归建立军用飞机价格预测模型。实例表明,在军用飞机价格预测方面,与原有的预测模型和逐步多元回归模型相比,应用PLSR预测的精度更高,更能体现采购价格与飞机性能参数之间的关系。

Application of Partial Least-squares Regression to the Acquisition Cost Estimating of Military Aircrafts

With the continual promotion of the performances of military aircrafts,various Cost Drive Factors make the former Acquisition Cost estimating model not be applicable for modern military aircrafts.Analyzing small samples and various Cost Drive Factors of military aircrafts' Acquisition Cost and considering the advantages of Partial Least-Squares Regression(PLSR) in analyzing multivariate data with small samples,an application of PLSR to military aircraft's Acquisition Cost estimating is proposed.Firstly,PLSR extracts the first and second principal components of military aircrafts' Acquisition Cost data to identify the outliers in samples;secondly,PLSR selects Cost Drive Factors with Variable Importance in Projection from the valid samples;finally,PLSR regresses to Cost Drive Factors and establishes the military aircraft's Acquisition Cost estimating model.The calculated results in the example show that,compared with the former estimating model and Stepwise Regression model,PLSR model has higher accuracy in military aircraft's Acquisition Cost estimating,and can reflect better the correlation between Acquisition Cost and performance parameters of military aircrafts.