Development of an AI Model for Predicting the Severity of Traffic Accidents Using Ensemble Techniques - Focusing on The Changwon Special City -

Abstract

In the air transport industry, the safe transportation of cargo using trucks is essential. This is because even if there are no issues during the air transport process, any problems with cargo transportation would prevent the air transport from functioning effectively. The Changwon Special City is in close proximity to international and military airports in neighboring regions such as Busan Metropolitan City and Sacheon City. The city is also home to key government agencies and companies in the aerospace industry, including the recently established Korea Aerospace Administration (KASA) and Korea Aerospace Industries (KAI), as well as over 2,500 major domestic defense contractors. This concentration has led to active cargo transportation related to the defense and aerospace industries within Changwon Special City. Due to the nature of freight transportation, where cargo trucks are large and carry multiple or heavy items, there is a high risk of serious injuries or fatalities in the event of a traffic accident. Therefore, it is crucial to predict areas with a high likelihood of accidents and take preventive measures in advance to minimize the occurrence of such incidents. This study leverages the industrial characteristics of Changwon Special City to identify the factors that most significantly impact the severity of cargo truck accidents. By focusing on key research questions (RQs), the study aims to predict areas where accidents are most likely to occur, thereby minimizing their occurrence. To achieve this, we analyze the characteristics of cargo truck accidents using data from incidents that occurred in Changwon Special City and develop an AI model to predict high-risk areas for accidents. To achieve this, an AI model will be used to comprehensively analyze freight truck accidents in Changwon Special Case City over a 10-year period from 2014 to 2023. The analysis will be divided into temporal factors such as day of the week, time, and season, and spatial factors such as accident location, road characteristics, and surrounding environment. By utilizing the number of casualties, the study will assess the severity of freight truck accidents in Changwon Special Case City and identify the most influential causes. Specifically, a correlation analysis will be conducted based on factors influencing the severity of traffic accidents, such as accident type and traffic law violations. An AI model will then be developed using ensemble techniques to predict traffic accident severity. At the same time, the road patterns in areas with frequent freight truck accidents will be analyzed to predict high-risk areas. In conclusion, the significance of this study is twofold, both academically and practically. Academically, this study makes a significant contribution by addressing the relatively underexplored area of the severity and prediction of cargo truck traffic accidents. Through this research, we advance the understanding of factors influencing cargo truck accidents and provide insights that were previously lacking in the existing literature. Second, the study utilizes an artificial intelligence model to analyze cargo truck traffic accidents and, through this analysis, implements a model that predicts the severity of such accidents. The practical significance of this study is as follows. First, local governments such as Changwon Special City and Gyeongsangnam-do, along with the police, can utilize the findings to design traffic accident prevention policies for high-risk areas within the region, thereby protecting the lives and property of local residents. Second, for businesses operating in Changwon Special City, the study’s insights can support the prevention of traffic accidents, contributing to the establishment of a stable logistics system and aiding in the management of production and inventory.