A Dynamic Multi-Stage Group Decision-Making Method for Credit Risk Monitoring Using Non-equidistant Interval Time Framework and Fuzzy Cognitive Prediction

Credit risk monitoring of defaulters is a multi-stage, dynamic process that requires continuous evaluation at policy-driven milestones. While traditional approaches predominantly rely on equidistant time intervals, they often fail to align with the irregular, event-based nature of regulatory decisions. To address this limitation, this paper proposes a novel decision-making method for credit risk monitoring by establishing a non-equidistant interval time frame. Unlike static or equidistant models, our framework allows for evaluation at flexible, key time points defined by practical management needs. Specifically, the methodology utilizes dynamic multigranularity linguistic tools to capture and preserve heterogeneous expert assessments over time and employs fuzzy cognitive maps to model complex causal relationships for predictive modeling, ensuring that the decision-making process is both forward-looking, robust, and information-rich. By shifting from equidistant time steps to policy-relevant intervals, the proposed method provides a more adaptive and interpretable monitoring mechanism. At each key time point, the framework integrates defaulters historical data, current data, and predicted future data, which are then aggregated using the TOPSIS method to compute credit risk scores and support entire-process monitoring. An illustrative example demonstrates that the framework effectively captures risk evolution during the monitoring process, offering superior decision-support value compared to traditional equidistant-interval methods.