Detection and Analysis of Anomalies in Municipal Expenditure Time Series

• Marco Túlio Dutra UFMG / UFOP
• Lucas G. L. Costa UFMG
• Gabriel P. Oliveira UFMG
• Mariana O. Silva UFMG
• Gisele L. Pappa UFMG

DOI: https://doi.org/10.5753/lasdigov.2025.8935

Abstract

This paper presents an approach for detecting and analyzing anomalies in time series of public expenditures. The approach employs a combination of advanced techniques for anomaly detection in time series, including statistical analysis and machine learning. In addition to anomaly detection, the approach allows the ranking of expenditures based on the number of detected anomalies and the monetary values involved, facilitating the prioritization of audits. Our approach is evaluated on a real dataset containing over one million municipal expenditure records from the state of Minas Gerais. Overall, our findings indicate that the approach effectively identifies and prioritizes cases with a high potential for irregularities.

References

Blázquez-García, A. et al. (2021). A review on outlier/anomaly detection in time series data. ACM Computing Surveys, 54(3):1–33.

Braz, C. S. et al. (2023). Análise de irregularidades em licitações públicas com foco em empresas de pequeno porte. In WCGE, pages 94–105. SBC.

Breunig, M. M. et al. (2000). Lof: identifying density-based local outliers. SIGMOD Rec., 29(2):93–104.

Costa, L. G. L. et al. (2024). Quanto Custa: Banco de Preços de Compras Públicas do Estado de Minas Gerais. In SBBD DS-CoPS. SBC.

Cuturi, M. et al. (2017). Soft-dtw: a differentiable loss function for time-series. In ICML, pages 894–903. PMLR.

Gomide, L. D. et al. (2023). Mineração de dados sobre despesas públicas de municípios mineiros para gerar alertas de fraudes. In SBBD, pages 378–383. SBC.

Handoko, B. L. et al. (2022). The effect of skepticism, big data analytics to financial fraud detection moderated by forensic accounting. In ICEEG, pages 59–66.

Hethu Avinash, D. et al. (2024). Integrating level shift anomaly detection for fault diagnosis of battery management system for lithium-ion batteries. IEEE Access, 12:116071–116084.

Hyndman, R. et al. (2018). Forecasting: Principles and Practice. OTexts, 2nd edition.

Liu, F. T. et al. (2008). Isolation forest. In ICDM, pages 413–422. IEEE Comp. Society.

Malhotra, P. et al. (2015). Long short term memory networks for anomaly detection in time series. In ESANN.

Mendes, B. M. A. et al. (2023). Impacto de Doações Eleitorais no Faturamento de Empresas: Um Estudo nas Eleições Municipais em Minas Gerais. In SBBD, pages 420–425. SBC.

Mongwe, W. T. et al. (2020). The efficacy of financial ratios for fraud detection using self organising maps. In SSCI, pages 1100–1106. IEEE.

Oliveira, F. B. et al. (2024). Machine learning and time series analysis to forecast hotel room prices. In BRACIS, pages 358–371. Springer.

Oliveira, G. P. et al. (2023a). Assessing data quality inconsistencies in brazilian governmental data. Journal of Information and Data Management, 14(1).

Oliveira, G. P. et al. (2023b). Ranqueamento de licitações públicas a partir de alertas de fraude. In BraSNAM, pages 1–12. SBC.

Silva, C. A. et al. (2020). Consciência, Prevenção e Detecção contra Fraude: Estudo com Auditores Internos e Contadores do Setor Público Brasileiro. In USP International Conference on Accounting.

Silva, M. O. et al. (2023). Análise de sobrepreço em itens de licitações públicas. In WCGE, pages 118–129. SBC.

Zamanzadeh Darban, Z. et al. (2024). Deep learning for time series anomaly detection: A survey. ACM Comput. Surv., 57(1).