Time-dependent rebar price prediction for procurement decision-making using bio-optimized deep machine learning

Purpose – This study aims to (1) develop an artificial intelligence (AI)-based model to accurately forecast rebar prices and (2) propose procurement strategies to reduce the subjectivity involved in rebar price trend forecasting and minimize procurement costs for construction project general contractors. Design/methodology/approach – Correlation analysis was used to identify the key factors influencing changes in rebar prices over time. An AI-based inference model, symbiotic bidirectional gated recurrent unit (SBiGRU), was developed for rebar price forecasting. The performance of SBiGRU was compared with other AI techniques, and procurement strategies based on the SBiGRU model were proposed. Findings – The SBiGRU model outperformed the other AI techniques in terms of rebar price forecasting accuracy. The proposed rebar price forecasting model (RPFM) and procurement patterns, which integrate inventory management principles and rebar price forecasts, were demonstrated to effectively optimize procurement costs, realizing a remarkable 6.13% reduction in procurement expenses compared to the conventional monthly procurement approach. Research limitations/implications – The accuracy of AI models may be impacted by disparities in the data used for model training. Future research should explore approaches incorporating price predictions and order factors. Originality/value – This study significantly extends the bounds of traditional rebar price prediction by integrating AI-driven forecasting with inventory management principles, highlighting the potential of AI-based models to improve construction industry procurement practices, reduce related risks and costs, optimize project operations and maximize project outcomes.