<![CDATA[This study aims to develop a classification model for the severity of workplace accidents using the Random Forest algorithm with OSHA–ITA data from January 2015 to September 2024. From 95,194 initial incident data, cleaning, missing value handling, removal of irrelevant variables, encoding, and the formation of Severity targets (Severe, Moderate, Mild) were carried out. The dataset was divided into 70% training data and 30% testing data with a total of 28,559 samples. The results showed that the initial model was biased towards the majority class, while the minority class was difficult to recognize. After applying SMOTE, the model's performance improved with more balanced predictions. The most influential features included Nature Title, Part of Body Title, Event Title, Source Title, Hospitalized, and Amputation. These findings emphasize the importance of selecting relevant features and data balancing techniques to improve the performance of Random Forest classification in occupational accidents.]]>
Copyrights © 2025
