Algoritmo en python para predecir y detectar anomalías en procesos contractuales de la alcaldía de Puerto Rico, meta, mediante técnicas avanzadas de machine learning

Abstract

The municipal government of Puerto Rico, Meta, manages a large volume of contractual processes each year, which are fundamental for institutional development and the efficient use of public resources. However, the variability in the amounts awarded and the lack of advanced analytical tools make budget planning and early detection of irregularities difficult. Currently, information analysis is limited to descriptive reports that do not allow for the anticipation of administrative or financial risks. In view of this situation, we propose the development of a Python algorithm based on machine learning techniques, aimed at predicting contract values and detecting anomalies in the contracting process. Time series models such as Prophet will be implemented to project trends and Isolation Forest to identify atypical behavior. In addition, the SHAP methodology will be incorporated to ensure the interpretability of the results. The proposal consists of six phases: data consolidation and cleaning, exploratory trend analysis, predictive modeling, anomaly detection, explainability, and integration of an interactive dashboard. The aim is to strengthen budget planning and efficiency in the contractual management of the Mayor's Office of Puerto Rico, Meta.