Quantifying civilian aircraft vulnerability: A data-driven approach in geo-political conflict zones for improved risk assessment of aircraft shot-down

Abstract

Current risk analysis methods for quantifying the risk of the shooting down of commercial aircraft rely on the use of risk matrices and risk categorisation classes. We show that these processes are not effective, subject to bias and not adequate to help aviation companies decide whether to fly to or over conflict areas.

Information concerning terror attacks, wars or conflicts is instantly available through various internet channels, and we argue that this enables more innovative accurate data-driven aircraft shoot down risk assessment. We propose a generalised linear model with logit link to estimate the likelihood of an aircraft being shot down based on technical and geo-political environmental factors. We use our model to estimate the probability of aircraft being shot down in all countries that are currently affected by military conflict. We demonstrate that probability of shooting down civilian aircraft depends on economic indicator such as GDP per capita, type and intensity of the conflict. We validate our model using out-of-sample tests with cross-validation.

The method proposed in this paper uses data available in open sources, it is easy to implement and utilize in aviation company or other industry bodies for prediction of aircraft shooting risks. It significantly improves currently existing methodologies of aircraft shooting risk assessment.