一种基于Logistic回归的全损飞行事故分析与预测

An Analysis and Forecast Model of Total Loss Flight Accidents Based on Logistic Regression Method

运用Logistic回归方法分析全损飞行事故的影响因素,并构建预测模型,然后利用飞行事故案例进行了实证检验。通过对样本数据过离散、空单元、多元共线性和特异值、特殊影响案例的诊断,以及对模型的拟合程度和预测效果的检验,表明所构建的模型比较符合现实情况,而且具有较好的拟合程度和预测效果。模型指出航班类型和机龄因素具有统计学意义,国内航班比国际航班的全损概率要高;机龄越老,全损概率越高。

Logistic regression method was applied to analyzing the affecting factors of total loss flight accidents,and on the basis of it,a forecast mode was built.Then the model was tested with some accident cases obtained by stratified random sampling from accident case database.Through overdispersion,zero cell count,multicollinearity,outliers,influential observations,fitting and predictive accuracy tests,the results demonstrate that the model is adequacy,effective and in accord with actuality.Meanwhile,the model indicates that flight type and aircraft age are of significance in statistics science,a domestic flight is more exposed to total loss than an international flight;the older an aircraft is,the higher its total loss probability is.