Optimizing Supply Chain Shipping Time Prediction: A Hybrid Approach Using a Modified Marine Predators Algorithm

Accurate Shipping Time Estimation (STE) remains a critical and complex challenge in supply chain management, requiring the application of sophisticated and flexible machine learning (ML) techniques. Precise shipping time predictions are essential for businesses seeking to improve operational efficiency and satisfy customer demands in dynamic, time-sensitive contexts. This study proposes a novel enhancement of the Marine Predators Algorithm (MPA), specifically designed for hyperparameter optimization of ML models targeting STE. Leveraging a publicly available shipping optimization challenge dataset, which provides a comprehensive and standardized benchmark, a hybrid modeling framework is developed by integrating ML algorithms with the enhanced optimization algorithm. This integration aims to elevate both the predictive accuracy and generalization capabilities of the models. The proposed approach is rigorously compared against the MPA and the existing versions modified for hyperparameter tuning. Experimental results reveal that the enhanced MPA-based model significantly outperforms competing methods, exhibiting superior robustness and estimation accuracy. The findings highlight the efficacy of the proposed enhancement in optimizing supply chain STE models and contribute meaningful insights to the advancement of supply chain analytics.