Time series prediction of macroscopic construction safety indicators using grey–Markov models

This study attempts to find a suitable method to predict the dynamic safety situation and future trend in the Chinese construction industry. Five types of grey–Markov models were examined to evaluate their predictive accuracy for the macroscopic selected construction safety indicators, specifically the accident rate per 1,000 million RMB and the death toll per 1,000 million RMB. The models tested include the grey Verhulst–Markov Model (GVMM), even grey–Markov model (EGMM), discrete grey–Markov model (DGMM), original difference grey–Markov model (ODGMM) and even difference grey–Markov model (EDGMM). Comparative analyses of the models’ predictive performance indicate that the GVMM significantly outperforms other alternative models in both safety indicators. The findings highlight the GVMM as a robust and effective tool for modeling and predicting dynamic safety conditions and future trends in the construction industry at a macro level.