Aviation safety analysis

Abstract

Aviation is a constantly growing industry in the modern world. With an average of 400 accidents within the industry occurring on a monthly basis, it is imperative to determine the causes of these accidents to improve aviation safety. In this work, data and text mining techniques are used to extract useful information from a large database of aviation accidents. The ASRS database, which consists of aviation accident reports since 1988, is used in this study. It consists of a detailed account of what occurred in the accidents, as well as categorical information about the flights in question, such as weather elements and aircraft information. The study of such accident reports helps to identify the reason for these accidents and extract similarities or differences amongst them, with the aim of preventing fatalities and the loss of resources. This work demonstrates the use of data mining techniques to determine the primary problem of accident reports from the ASRS and predict the risk factor of these accidents. This is achieved through the use of machine learning classifiers such as Naive Bayes, SVM and Random Forest, and deep learning techniques. From the experiments evaluated, the best performing method for identifying the primary problem is the MLP classifier on structured data. Furthermore, for the prediction of risk factor, SVM provided the best results on structured data. Although the results of structured data were better than the unstructured data, our manually created dictionaries, used to reduce the vocabulary in the texts, improved results on the unstructured data. The work achieved with our system demonstrates that machines can reliably identify a flight’s primary problem, as well as a high-risk situation in a flight