FP-Growth-based risk pattern discovery for dual cost-risk mitigation in resilient multi-sourcing order allocation under time-varying demand

The huge growth of e-commerce and globalization has increased the complexity of inbound supply chain resilience, especially in multi-sourcing order allocation decisions. Despite extensive research, traditional methods often prove inadequate for handling the dual cost-risk minimization, particularly when confronted with dynamic changes. This research aims to fill this gap by proposing a novel solution that simultaneously tackles the dual objectives with order allocation under time-varying demand. The uniqueness of our approach lies in the application of the FP-Growth association-rules algorithm to uncover latent risk patterns based on criteria interdependency. We hybridize this technique with a new proposed variant of the Weighted Sum Method (WSM), offering a risk-aware decision-making model. A numerical application based on real-world case study from the automotive industry is proposed to demonstrate the applicability. Furthermore, to adapt to data availability constraints, we proposed a data augmentation algorithm using the Joint and Conditional Probability Distributions (JCPD) method. This technique generates volumetric synthetic data while preserving interdependencies between attributes (criteria), enabling realistic validation. Experimental results demonstrate the effectiveness of the proposed solution across various scenarios, highlighting its superiority. This research provides twofold theoretical and managerial implications. First, it introduces a theoretical understanding of a bottom-up risk-pattern discovery approach for real-time risk modelling in an AI-driven model. Second, it introduces a practical AI-driven stepwise approach for purchasers that enhances cost efficiency, resilience, and proactive risk mitigation in the emerging risk-prone markets. Despite the promising results, we acknowledge the limitations of our study and suggest future research directions.