Analysis of Commercial Airplane Accidents Worldwide Using K-Means Clustering

• Published in: International Journal of Safety and Security Engineering
• Main Author: Passarella R.; Iqbal M.D.; Buchari M.A.; Veny H.
• Format: Article
• Language: English
• Published: International Information and Engineering Technology Association 2023
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85178146986&doi=10.18280%2fijsse.130505&partnerID=40&md5=d49c3353f8c34f3c0ec0f3f49eec783f

Despite the Bureau of Transportation Statistics affirming the relative safety of air travel, with the lowest annual accident rate among various transportation modes, the importance of analyzing and mitigating aviation accidents remains paramount for the sustained safety and comfort of air travelers. This study leverages data from the Bureau of Aircraft Accident Archives (BAAA-acro) website, transformed into a dataset that encapsulates commercial airplane accident data spanning the period from 1918 to 2020. The dataset, comprising 110 observations across four variables, was subjected to K-means clustering to categorize the causes of airplane accidents. The optimal number of clusters for this analysis was determined using the Silhouette index. The investigation focused on two accident groups within the dataset. The first cluster, consisting of 106 observations, demonstrated a considerable degree of heterogeneity, indicative of a broad distribution and significant variation. The second cluster, comparatively smaller, comprised only four observations. The clustering exercise underscored that technical factors predominantly contribute to commercial airplane accidents. The findings of this study thus suggest that future efforts by aviation regulatory bodies to decrease aviation accident occurrences could benefit significantly from a concerted focus on these technical factors. © 2023 WITPress. All rights reserved.