Abstracts Track 2024


Area 1 - Internet Technology

Nr: 51
Title:

An Innovative and Cost-Effective Approach to Fraud Detection in Public Procurement Using Data Analysis and Automation Tools in Developing Countries

Authors:

Daniel de Sousa Camacho, Roberto Baptista Junior, Claudio Wang and Thiago Viegas Frossard

Abstract: Corruption in public procurement undermines public trust and diverts resources. Combating this challenge requires innovative and cost-effective approaches. This study presents a solution that integrates Python for data analysis with Microsoft’s low-code tools, Power Automate and Power BI, to enhance transparency and detect fraud in bidding processes. The focus is on identifying irregularities using open-source and existing government tools, reducing financial barriers. The methodology involves using Brazil's Painel Nacional de Contratações Públicas (PNCP) API to collect procurement data, which is processed in Python. The tool analyzes data to detect anomalies such as abnormally high bids or errors in data entries by public agencies. Python's Pandas library is used to clean and normalize data for structured analysis. Automation is integrated through Power Automate, sending real-time alerts to the audit department when risks are detected, enabling timely action. Power BI provides interactive dashboards that allow stakeholders to explore procurement data and spot risks using heatmaps and trends. These visualizations help identify breaches of spending limits and other issues. The system enhances transparency, improves auditing efficiency, and reduces the manual effort needed for fraud detection by enabling auditors to focus on high-risk cases. The study tested this solution using procurement data from São Paulo. It revealed several key anomalies, such as vendors winning contracts with unusually low bids, rapid contract awards, and vendor concentration in specific categories. These findings led to further investigations, resulting in the suspension of certain contracts. The tool's impact on identifying high-risk areas highlights its potential to improve public procurement processes. This solution is particularly beneficial for developing countries where financial and technological resources are limited. By leveraging Python, an open-source tool, and using cost-effective platforms like Power Automate and Power BI—already under contract with the São Paulo government—the system avoids significant additional expenses. Its scalability allows for the monitoring of growing procurement data volumes while remaining adaptable to different governmental contexts. The integration of Python for data processing with low-code automation and visualization significantly enhances fraud detection in public procurement. Future work could include predictive analytics to forecast risks based on trends and refined machine learning models for increased precision. This study highlights the importance of accessible technological solutions in addressing the issue of procurement fraud. Combining Python with Microsoft’s low-code platforms provides a cost-effective, scalable system that improves transparency and fosters accountability in public sector operations.