Forecasting aircraft mishaps using monthly maintenance reports

Abstract

Naval Aviation aircraft mishaps continue to be of great concern due to the high cost of loss of life and aircraft. The goal of this thesis is to develop a predictive statistical model that accurately forecasts Marine Corps AV-8B Harrier aircraft mishaps based on existing monthly maintenance reports. Monthly maintenance reports provide numerous independent variables based on personnel levels and maintenance hours that could possibly be used to forecast aircraft mishaps. These variables were graphically analyzed to determine any relationships that could be exploited in developing the model. Higher order relationships were investigated by the method of principal components and logistic regression. After a thorough analysis, there appears to be no combination of variables in this particular data that could be used to forecast aircraft mishaps. The overall result of the thesis is that there is no relationship between monthly maintenance reports and aircraft mishaps that can be exploited to develop a predictive statistical model