Analysis and Predictive Modeling of Stall and Spin Accidents in General Aviation Fixed-Wing Aircraft
Author ORCID Identifier:
Date of Graduation
8-2025
Document Type
Thesis
Degree Name
Master of Science in Mechanical Engineering (MSME)
Degree Level
Graduate
Department
Mechanical Engineering
Advisor/Mentor
Majumdar, Neelakshi
Committee Member
Campbell, Jenn
Second Committee Member
Huang, Po-Hao Adam
Keywords
Accident Severity Analysis; Aerospace Engineering; Fixed-wing Aircraft; General Aviation Safety; Predictive Modeling; Stall and Spin
Abstract
Stall and spin accidents represent a persistent safety concern in General Aviation, with approximately 41% resulting in fatalities despite their relatively low occurrence. This study analyzes 1,079 General Aviation stall and spin accidents from 2008 to 2023 using the National Transportation Safety Board accident database. Our analysis identified the top phases of flight, injury outcomes, and causal factors in stall/spin accidents. Findings indicate that over 93% of stall/spin events involved General Aviation aircraft, with most accidents occurring during initial climb, takeoff, or low-altitude maneuvering. Pilot-related control errors, especially poor airspeed and angle-of-attack management, were the most common causes. To assess accident severity prediction, we trained machine learning models—Logistic Regression, Decision Tree, Random Forest, and XGBoost—on structured features. The Random Forest model achieved the best performance (accuracy = 0.81, ROC-AUC = 0.89), demonstrating potential for integration into safety monitoring and training systems. The findings offer an interpretable, data-driven framework for identifying high-risk scenarios and informing targeted training to prevent stall/spin accidents in General Aviation. Building on prior text-mining and state-based modeling of NTSB reports, we also examined environmental factors such as airport elevation, altimeter settings, and light conditions, as well as human factors like pilot age. Feature-importance analysis revealed that in-flight loss of control, engine power loss, and recovery attempts are among the strongest predictors of fatal outcomes. Temporal trends show a modest decline in fatality proportion since 2015, suggesting improvements in pilot training and aircraft safety systems. The resulting severity-prediction tool is highly interpretable and can be embedded in real-time flight-data monitoring platforms to support proactive risk mitigation in General Aviation.
Citation
Hasan, M. (2025). Analysis and Predictive Modeling of Stall and Spin Accidents in General Aviation Fixed-Wing Aircraft. Graduate Theses and Dissertations Retrieved from https://scholarworks.uark.edu/etd/5916
