Intelligent systems for prediction, clustering, time series analysis, and legal judgment outcome modeling applied to energy sector and public administration data. Research spans Balanced Bagging classifiers, XGBoost, gradient boosting, and deep learning architectures applied to real-world large-scale datasets.
The Artificial Intelligence & Data Science research line at NCA-UFMA develops intelligent computational systems capable of extracting insights from complex, high-volume, and heterogeneous data sources. Our work addresses real-world challenges across the energy sector, public administration, healthcare, and logistics.
Key research threads include unsupervised customer dissatisfaction detection in electric utility data, legal judgment outcome prediction for consumer lawsuits using ensemble models, time series prediction and imputation for energy consumption, spatial clustering of residential consumers, and data-driven public management dashboards.
A flagship project analyzed 2.5 million customers of Equatorial Energia, achieving 96.52% sensitivity in detecting unreported power consumption (UPC) using Balanced Bagging over spatial, financial, and consumption features from the preceding 18 months.
CHESF / ANEEL P&D
Integrated GIS-based platform for CHESF's electrical system covering SCADA/EMS supervision, alarm management, power quality, oscillography, environmental licensing (reservoir borders, PAS), and lightning/wildfire event correlation. Developed in partnership with UFCG.
Equatorial Energia / ANEEL P&D
Intelligent system for continuous detection of dissatisfied customers, identification of dissatisfaction drivers, and suggestion of pre-litigation agreements. Integrates the virtual assistant Clara for proactive consumer outreach via WhatsApp.
Equatorial Energia / ANEEL P&D
Computational intelligence system that identifies consumer profiles predisposed to filing legal actions against the energy distributor, using historical records from CP-PRO, SAP-CCS, SAP-CRM, and judicial databases to prioritize preventive action.
SEPLAN-MA / São Luís City Hall / IMESC
Suite of GIS-based socioeconomic indicator monitoring platforms for state and municipal governments, including the SEPLAN PPA dashboard, the São Luís municipal indicators system (DIEE), and the Maranhão municipal indicators platform for IMESC.
TCE-Maranhão (Maranhão State Court of Accounts)
Two integrated systems for the Maranhão State Court of Accounts: E-PCA (Electronic Public Accounts Submission) enabling jurisdictioned entities to file annual accounts digitally, and META (Auditor Workbench) providing data science-based audit analytics with interactive dashboards.