Ver o conteúdo principal

Data Science for the Evaluation of Operational Errors and Their Impact

Projetos T

About the Project

Data Science for the Evaluation of Operational Errors and Their Impact

The project aims to develop new methodologies based on data science, including artificial intelligence, to estimate and predict the extent of errors and their impacts on operational programs supported by EU community funds in Portugal. The goal is to utilize these innovative methodologies as a replacement or complement to traditional audit practices, which rely solely on sample-based audits.

In Portugal, thousands of audit activities are conducted annually for funded operations, requiring significant time, human resources, and costs. Errors estimated at the program level have diverse implications, such as financial corrections (including extrapolated error corrections), the application of flat-rate corrections by the European Commission, and the need to develop improvement plans. Given the substantial impact of these errors on public policies and the significant financial volumes involved in these corrections, it is crucial to ensure the reliability of the estimates while simultaneously simplifying the effort required, especially in terms of human resources.

Traditional sample-based audits can sometimes yield inconclusive results, necessitating additional audit work. In recent years, the coordinator of this proposal has worked as an expert for the European Commission, developing innovative sampling methods—such as new standard monetary unit sampling, subsampling techniques, and multi-period sampling methods—which have significantly reduced audit efforts without compromising the confidence levels in error estimates. Despite these advancements, further innovation is needed to address the growing complexities of operational audits.

Our Contribution

NOVA IMS coordinated this project and was responsible for designing and implementing innovative analytical models. NOVA IMS focuses on developing machine learning algorithms and statistical methods to enhance the accuracy and reliability of error estimates. Specific contributions include:

  • Predictive Modeling: Building machine learning models to forecast potential operational errors, integrating diverse data sources for comprehensive error assessments.
  • Impact Analysis: Developing tools to quantify the financial and operational impacts of identified errors on EU-funded programs.
  • Efficiency Optimization: Implementing methods to complement or replace traditional sampling-based audits, reducing human and financial resource requirements while maintaining high confidence levels.
  • AI-Driven Decision Support: Designing decision support tools that utilize AI for identifying patterns and providing actionable insights for program improvement.

Through its contributions, NOVA IMS ensures the project's methodologies are both scientifically robust and practically applicable, enhancing operational efficiency in the audit process.

Funding

Funding programme: Agência para o Desenvolvimento e Coesão, IP - Programa Operacional Assistência Técnica (POAT)

Funding to NOVA IMS: €139 211,29

Duration: 2021 - 2022

Contribution to the SDGs