Scholarlink

<p>Business intelligence and data-driven segmentation increasingly support service prioritization and operational-risk monitoring in enterprise resource planning financial workflows. This study examines an SAP cross-company check-payment setting in which centralized treasury executes payments correctly, but fragmented payment-traceability across company-code documents creates repeated vendor follow-ups and additional customer-care workload. The study combines design-based process verification with survey-based, associational segmentation modelling. A traceability enhancement was evaluated in a quality-assurance sandbox by synchronizing check-reference visibility across relevant payment documents without changing accounting postings, clearing logic, or document-integrity controls. The behavioural component used 120 complete vendor responses collected through a five-point Likert instrument, with high follow-up burden defined as a survey-derived binary outcome. An interpretable logistic classifier was used as the primary auditable model and benchmarked against rule-based scoring, decision tree, random forest, and gradient boosting baselines. The logistic model achieved an area under the receiver-operating-characteristic curve of 0.75, precision–recall area of 0.61, F1-score of 0.66, and balanced accuracy of 0.72. After calibration, it achieved a Brier score of 0.18 and an expected calibration error of 0.03. Gradient boosting achieved higher discrimination, with an area under the receiver-operating-characteristic curve of 0.82, indicating an explicit trade-off between predictive discrimination and auditability. Cohort-level disparity reporting showed small observed tenure and vendor-type gaps, but these results are interpreted as descriptive monitoring rather than confirmatory fairness testing because of the modest sample size and 36 positive cases. Because predictors and outcome were collected from the same respondents at one time point, the behavioural findings are interpreted as associational. The findings support the feasibility of a scoped approach to auditable segmentation and traceability monitoring in SAP cross-company check-payment support operations.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Explainable business intelligence for auditable vendor segmentation and payment-traceability monitoring in SAP cross-company check payments

  • 期刊论文

  • 发表日期：

    2026年6月30日

查看全文
  • Sanjib Kumar Mishra,
  • Sasmita Mishra,
  • Rojalina Priyadarshini,
  • Zefree Lazarus Mayaluri

摘要

Business intelligence and data-driven segmentation increasingly support service prioritization and operational-risk monitoring in enterprise resource planning financial workflows. This study examines an SAP cross-company check-payment setting in which centralized treasury executes payments correctly, but fragmented payment-traceability across company-code documents creates repeated vendor follow-ups and additional customer-care workload. The study combines design-based process verification with survey-based, associational segmentation modelling. A traceability enhancement was evaluated in a quality-assurance sandbox by synchronizing check-reference visibility across relevant payment documents without changing accounting postings, clearing logic, or document-integrity controls. The behavioural component used 120 complete vendor responses collected through a five-point Likert instrument, with high follow-up burden defined as a survey-derived binary outcome. An interpretable logistic classifier was used as the primary auditable model and benchmarked against rule-based scoring, decision tree, random forest, and gradient boosting baselines. The logistic model achieved an area under the receiver-operating-characteristic curve of 0.75, precision–recall area of 0.61, F1-score of 0.66, and balanced accuracy of 0.72. After calibration, it achieved a Brier score of 0.18 and an expected calibration error of 0.03. Gradient boosting achieved higher discrimination, with an area under the receiver-operating-characteristic curve of 0.82, indicating an explicit trade-off between predictive discrimination and auditability. Cohort-level disparity reporting showed small observed tenure and vendor-type gaps, but these results are interpreted as descriptive monitoring rather than confirmatory fairness testing because of the modest sample size and 36 positive cases. Because predictors and outcome were collected from the same respondents at one time point, the behavioural findings are interpreted as associational. The findings support the feasibility of a scoped approach to auditable segmentation and traceability monitoring in SAP cross-company check-payment support operations.