基于支持向量机的民航飞行人员人格选拔分类器的探讨

目的:探究将统计学习方法应用于心理测验所得的大量数据进行学习分析的可行性,并基于探究结果对飞行职业的人格特征进行进一步探索,为飞行人员的选拔及评估提供新的思路。方法:从某航空公司随机抽取1020名男性被试,其中飞行人员510名,非飞行人员510名,采用卡特尔16项人格测试对其进行测验,施测后对得到的16项因子分采用支持向量机就随机划分的训练组和测试组进行学习,分析学习结果。结果:挑选出4项因子作为分类的特征因子,基于线性支持向量机构建的分类器在交叉验证下的平均正确率为64%。结论:采用SVM构建的分类器具有一定的可靠性和有效性。

A Pilot Study of Assessment of Civilian Pilot Personality based on Support Vector Machine

ABSTRACT Objective: To explore the feasibility of using statistical learning method to mine the mass data obtained from psycho- logical tests by using the airline employees'' 16PF data to construct a classifier based on SVM and expose the distinctive competency characters of civilian pilot, which will provide a new way for the personality selection and evaluation of civilian pilot. Methods: 1020 em- ployees, 510 pilots and 510 others, sampled by random were investigated with 16 personality factor questionnaire (16PF). A selection and evaluation system is constructed based on learning the standardized 16 personality factor scores by support vector machine (SVM). Results: Four factors are chosen as feature factors, which are emotional stability, sensitivity, abstractedness and perfectionism. The cross-valida- tion error score of the classifier constructed based on linear SVM is 64%. Conclusion: Simulation example shows that the proposed method is effective, reliable and practical.