Applying data mining techniques to explore factors contributing to occupational injuries in Taiwan's construction...

Area of Science:

• Occupational Safety and Health
• Construction Management
• Data Mining in Safety Research

Background:

• Construction industry faces significant occupational hazards, including injuries and fatalities.
• Systematic analysis of accident databases is crucial for understanding and mitigating risks.
• Taiwan's construction sector has a substantial record of occupational accidents requiring in-depth investigation.

Purpose of the Study:

• To explore the causes and distribution of occupational accidents in Taiwan's construction industry.
• To establish potential cause-and-effect relationships for serious construction accidents.
• To provide a data-driven framework for enhancing construction safety practices and training.

Main Methods:

• Utilized a comprehensive database of 1542 construction accident cases from 2000-2009.
• Applied the data mining technique Classification and Regression Tree (CART) for analysis.
• Systematically sorted, classified, and encoded injury and fatality data.

Main Results:

• Identified specific occurrence rules for falls and collapses as critical predictors of occupational injuries.
• Demonstrated the effectiveness of CART in uncovering significant cause-and-effect relationships in accident data.
• Found that falls and collapses are key factors in both public and private construction projects.

Conclusions:

• Falls and collapses are primary indicators for predicting occupational injuries in construction.
• The study provides actionable insights for improving safety protocols and worker training programs.
• Findings contribute to a proactive approach to preventing construction accidents and protecting workers.