The industrial sector in South Korea has recently undergone significant growth; however, it is also known for its hazardous workplaces. Occupational accidents have had a widespread impact across various industries; therefore, the identification of accident-influencing factors is crucial to improve workplace safety. We analyzed the occupational accident database from the Ministry of Economy and Finance to examine the influencing factors, including worker information, project details, time-related variables, and accident descriptions. Exploratory and correspondence data analyses were performed to identify patterns and relationships between variables. We applied multinomial logistic models and random forest algorithms to understand the correlation between victim status and independent variables. Results showed that 67% of all accidents occurred among workers with less than one month of employment. The multinomial regression model achieved a prediction accuracy of 97.66% with a kappa value of 0.846, outperforming the random forest model (kappa = 0.844). The receiver operating curve illustrated that the random forest had higher misclassification rates when distinguishing between injuries and fatalities. To mitigate accidents among new workers, enhanced safety training and protective measures are needed to enforce a healthy workplace. This study contributes to ongoing efforts to advance workplace safety, reduce occupational accidents, and promote a healthier working environment.