Integración del algoritmo K-means en PostgreSQL mediante extensiones en PL/Python: una arquitectura medianamente acoplada
Year: 2025
Type: article
Abstract: This paper presents the process of integrating the K-means partitioning clustering algorithm into the PostgreSQL data-base management system by developing an extension in PL/Python. The proposal is framed within a moderately coupled architecture, aiming to incorporate machine learning capabilities directly into the DBMS. This integration allows clustering processes to be executed without the need to rely on external tools, thus facilitating access to data mining techniques for small and medium-sized organizations and contributing to the generation of knowledge from large volumes of data, with a view to improving their competitiveness. The research is focused on extending PostgreSQL by incorporating the descriptive technique of clustering, specifically the K-means algorithm, to segment datasets into homogeneous groups. To achieve this, the CRISP-DM (Cross-Industry Standard Process for Data Mining) methodology was adopted, which is suitable for data mining projects and comprises six phases: business understanding, data understanding, data preparation, modeling, evaluation, and deployment. As a result, an extension was designed, developed, and implemented to expand PostgreSQL's functionalities that made it possible to execute clustering processes efficiently. The solution was validated through comparative tests against specialized tools such as Weka and KNIME, showing consistent and high-quality results in data grouping. The extension maintains lower response times compared to WEKA and KNIME, and its performance is optimal for high data volumes and clustering exercises of greater dimensional complexity. In conclusion, the developed extension demonstrates that it is possible to enhance PostgreSQL's analytical capabilities through the integration of machine learning algorithms. This proposal represents a viable and accessible alternative for organizations that need to perform advanced data analysis without depending on external platforms.
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7154686303