Business Intelligence em dados eleitorais: uma Arquitetura baseada em Esquema Constelação

Abstract

This technological report aimed to propose and evaluate the application of the Galaxy Schema architecture for consolidating and integrating Brazilian public electoral data from the 2022 elections, demonstrating its viability as a foundation for decision-support systems. The problem addressed is the non-standardized and unintegrated release of electoral, demographic, and territorial data by various Brazilian public agencies—each using distinct formats and levels of granularity. This situation necessitates lengthy, complex ETL processes for cross-referencing information and leaves the average citizen dependent on intermediaries to interpret public data, thereby compromising their analytical autonomy. The proposed solution involved implementing an ETL pipeline to clean and standardize geographic codes and municipal names, followed by modeling a Data Warehouse using the Galaxy Schema. This structure utilized three shared dimensions—location, time, and demographic profile—to interconnect multiple thematic fact tables (population, territory, education, and health) without attribute duplication. Additionally, interactive analytical dashboards were developed in Tableau at various geographic levels (national, state, and municipal, with a focus on Uberlândia). As an achieved result, the electoral, demographic, and territorial databases were consolidated into a Data Warehouse structured under the Galaxy Schema, on top of which interactive analytical dashboards were developed for exploring electoral data at different geographic levels (national, state, and municipal). As expected results, once the dashboards that cross sociodemographic and electoral data are completed, these include enabling direct, agile cross-referencing between these datasets, the ability to identify relationships between sociodemographic conditions (income, sanitation, education) and voting behavior by region, and the potential to reveal previously unidentified patterns that could inform future studies and public policy discussions.