Machine learning methods have gained widespread utilization in small and micro enterprise credit risk assessment. However, the practical application of these methods encounters a conundrum involving accuracy and interpretability. In this study, a multi-stage ensemble model is proposed to enhance the model’s interpretability. To strengthen predictive portraits, a multi-feature enhancement method is proposed to integrate non-financial behavioral information and soft information on credit rating into the annual loan ledger data, thereby bolstering the explanatory capacity of the features. To rectify the issue of data imbalance and avoid information loss, a new bagging-based oversampling method is proposed to oversample the minority class samples in multiple parallelized subsets divided by the bagging strategy. To unleash the performance potential of base classifiers, a new voting-weight optimization method is proposed to optimize the soft voting weights of the candidate base classifiers. The experiment results of an annual loan ledger dataset of a commercial bank in China (with an accuracy of 97.9%, an area under the curve of 0.97, a logistic loss of 0.07, a Brier score of 0.01, and a Kolmogorov-Smirnov statistic of 0.38) and the other five public datasets indicating excellent model fit. By focusing on the widespread soft information and data structures characteristic of SME loan risk assessment data, an additional SHAP model explanation method enhances interpretability. This method reveals that the enhanced ’debt-to-income ratio,’ along with non-financial behavioral information and features derived from soft information, are essential for predicting loan defaults. Such enhancements help to alleviate the issue of information asymmetry in SME loan risk assessment.

We introduce an evolving network model of credit risk contagion in the credit risk transfer (CRT) market. The model considers the spillover effects of infected investors, behaviors of investors and regulators, emotional disturbance of investors, market noise, and CRT network structure on credit risk contagion. We use theoretical analysis and numerical simulation to describe the influence and active mechanism of the same spillover effects in the CRT market. We also assess the reciprocal effects of market noises, risk preference of investors, and supervisor strength of financial market regulators on credit risk contagion. This model contributes to the explicit investigation of the connection between the factors of market behavior and network structure. It also provides a theoretical framework for considering credit risk contagion in an evolving network context, which is greatly relevant for credit risk management.

From the interdisciplinary perspective of industrial economics and behavioral finance, this study establishes a noncooperative-cooperative biform game model between new energy enterprises and traditional energy enterprises. In this model, sales price is considered the non-cooperative strategy, while R&D expenses borne forms the basis of cooperative alliances. The Shapley value is applied to allocate profits, and numerical analysis is conducted to analyze the impact of factors, such as government subsidies and competitive intensity, on optimal strategies. The findings reveal the following: (1) Government subsidies effectively increase energy sales volume, promote technological advancements in new energy enterprises, and reduce the traditional energy enterprises’ proportion of R&D expenses borne. Moderate increases in competitive intensity are conducive to expanding market size, thereby enhancing both energy sales volume and profits. (2) Reasonably increasing the executive risk preference of energy enterprises encourages traditional energy enterprises to bear a higher proportion of R&D expenses and stimulates new energy enterprises to improve their production level, leading to increased sales value of energy. (3) Rising investment and production costs result in a higher proportion of R&D expenses borne for new energy enterprises. Consequently, the shrinking of new energy value reduces their profits, while the profits of traditional energy enterprises increase correspondingly.

Process anomaly detection in financial services systems is crucial for operational compliance and risk management. However, traditional process mining techniques frequently neglect the detection of significant low-frequency abnormalities due to their dependence on frequency and the inadequate incorporation of domain-specific knowledge. Therefore, we develop an enhanced process mining algorithm by incorporating a domain-specific follow-relationship matrix derived from standard operating procedures (SOPs). We empirically evaluated the effectiveness of the proposed algorithm based on real-world event logs from a corporate account-opening process conducted from January to December 2022 in a Chinese commercial bank. Additionally, we employed large language models (LLMs) for root cause analysis and process optimization recommendations. The empirical results demonstrate that the E-Heuristic Miner significantly outperforms traditional machine learning methods and process mining algorithms in process anomaly detection. Furthermore, the integration of LLMs provides promising capabilities in semantic reasoning and offers explainable optimization suggestions, enhancing decision-making support in complex financial scenarios. Our study significantly improves the precision of process anomaly detection in financial contexts by incorporating banking-specific domain knowledge into process mining algorithms. Meanwhile, it extends theoretical boundaries and the practical applicability of process mining in intelligent, semantic-aware financial service management.

The efficiency and reliability of business processes in commercial banks are critical to financial stability and compliance. However, traditional evaluation methods that rely on retrospective qualitative assessments and static frameworks struggle to address the dynamic complexities inherent in modern banking operations. These approaches lack real-time monitoring, fail to leverage granular event log data, and overlook organizational interdependencies, hindering proactive risk management and optimization. To bridge these gaps, this study proposes a data-driven evaluation framework that integrates three core dimensions: efficiency, quality, and flexibility. We developed a hybrid analytical model by integrating process mining with DEMATEL-AHP to analyze a Chinese bank’s performance guarantee process, comparing pre- and post-centralization workflows. The analysis revealed that post-centralization processes exhibited improved flexibility but reductions in efficiency and quality. Moreover, the social network analysis highlighted structural shifts, including expanded audit participation and reduced departmental cohesion, contributing to inefficiencies. This study advances business process management by demonstrating that a data-driven process evaluation framework offers greater persuasiveness and methodological rigor than traditional qualitative approaches.