DAVID ANTONIO FRANCO BORRÉ,RAÚL JOSÉ MARTELO GÓMEZ,DEIMER ANTONIO ROMERO MADERA
DOI: https://doi.org/This study presents an analysis of the impact of machine learning on modern accounting by studying the ContaWeb-BI platform, a solution jointly developed by the University of Cartagena and Colciencias to strengthen accounting management and strategic decision-making. The predictive and classification models implemented are described, and their ability to generate key performance indicators useful in strategic decision-making is evaluated. An exploratory methodology is adopted based on the ContaWeb-BI case study. Linear regression, decision trees, and K-means clustering techniques were applied to project sales, detect potential financial fraud, and segment suppliers, respectively. Empirical validation was performed with simulated data in an operational functional environment, and the results showed a high level of accuracy in all models: an R² of 0.88 in regression, an F1 score of 0.83 in classification, and a Silhouette coefficient of 0.69 in cluster analysis. The models were integrated into the platform's backend and deployed in an interactive dashboard that automates the generation of key performance indicators. This work demonstrates the technical and practical feasibility of using machine learning algorithms in accounting environments and proposes a replicable model for other organizations interested in adopting accessible, interpretable, and action-oriented analytical technologies.
