Temporal Analysis of Daily Fatalities Using K-Means Clustering Method

Abstract

Malaysians make long-distance travel during festive seasons, when traffic increases by approximately 30%. The high exposure on the roads inevitably increases the risk of traffic crashes. By using daily traffic fatalities between 18th March 2025 to 25th May 2025, this study examines the effect of different types of holidays on traffic fatalities. Holidays were classified into three categories: school holidays, public holidays, and seasonal holidays. The K-Means clustering method was employed for this analysis, and the results revealed that the type of holiday played a significant role in influencing the traffic fatality patterns. Three distinct clusters were deduced in this analysis. The highest risks took place in early June during the seasonal holidays, while the lowest fatalities, with an average of 12.22 fatalities, were observed around September. Moderate risks were reported in the mid of March 2025. Understanding the effect of seasonal variation on traffic fatalities enables relevant government agencies to develop targeted interventions during the respective seasons. As this study only considered limited variables such as day of week and type of holiday, future research can incorporate additional factors such as weather conditions and potential spatial correlations to enhance the predictive insights.