Publication

Abstract : Workplace safety is always a concern of utmost importance in any organization. Studies have shown that the fatality rate is highest in the construction industry among all other industries. The construction project managers need to understand the risk status of each of their projects and thus implement preventive measures. The introduction of digital tools into construction sites not only reduces the health and safety hazards among workers but also paves the way to the economic growth of the industry. The present study implements techniques of data analytics and Machine Learning (ML) into the construction safety sector. For the analysis, a dataset with 4847 incident reports during 2015–2017 from Occupational Safety and Health Administration (OSHA) database is used. Initially, the major attributes contributing to the incident are identified. Based on these identified factors, they were classified as Before Accident and After Accident attributes, and ML algorithms are used for the prediction of the construction fatality. Performance evaluation of these ML algorithms shows us that Random Forest (RF) has better prediction results for Before Accident attributes, whereas Decision Tree (DT) performed well for After Accident attributes. From a broader perspective, this study will help the safety management team to understand the severity of safety risks faced in each of their construction projects and also facilitate the implementation of proper preventive safety mechanisms.