Influencing Factors of Workers’ Unsafe Behaviors in the Construction Cycle of Commercial Building: A Dual Perspective of Frequency and Entropy

Unsafe behaviors by construction workers are a primary cause of accidents in commercial building construction. While traditional studies focus on the frequency of violations, they often overlook the disorder and unpredictability of such behaviors. This study introduces “Unsafe Behavior Entropy” as a new index to measure the disorder of workers’ behaviors, complementing traditional violation frequency. Utilizing a dataset from a large-scale commercial building construction project in Wuhan, China, this research uses Partial Least Squares Regression (PLSR) and Gray Relational Analysis (GRA) to examine the influence of six key factors, including safety meeting coverage and supervision density. The PLSR results indicate that the number of workers supervised per safety officer is the most critical driver of both frequency and entropy, while the coverage rate of entry safety education significantly impacts behavioral stability. GRA findings further reveal a high degree of correlation between management interventions and reductions in behavioral disorder. The study concludes that optimizing safety resource allocation and standardizing educational processes are fundamental to controlling human-related risks. By integrating the dual perspectives of frequency and entropy, this research provides a more comprehensive framework for safety management in complex building projects.