Understanding Incident Trends: A Deep Dive Into Train Incidents

This study investigates the factors leading to the classification of locomotives as long-standing (out of service for over 50 days) in South Africa. It focuses on trends in train incidents, identifying 363 longstanding locomotives with unscheduled maintenance, collisions, and vandalism listed as the primary incident types. Time series decomposition uncovered a seasonal pattern, and correlation analysis showed significant relationships among these incident categories. The Prophet model was selected for forecasting, predicting an increase in incidents related to maintenance, collisions, and vandalism. The study emphasises the need for further research on the impact of incidents on safety, operations, and personnel, as well as the effectiveness of safety training programmes. The findings stress the importance of understanding incident trends and their implications for railway safety and operational efficiency, providing insights for strategies to mitigate the impact of train incidents in South Africa.