Data analytics model to predict possible disciplinary sanctions against officials who hold popularly elected positions in Colombia

Abstract

This article describes the design of a data analytics model to predict possible disciplinary sanctions against officials who hold popularly elected positions in Colombia, a data exploration process was carried out and different Machine Learning models were applied in order to determine which of them is the most suitable for this purpose. Which in turn made it possible to identify risk situations and anticipate them through the implementation of preventive measures. To this end, a supervised learning approach was applied in the construction of the model, in which classification models were used to predict whether a given staff member might be subject to disciplinary sanctions in the future. One of the key aspects of this article was the optimization of the model's hyperparameters, as good accuracy and optimal performance were achieved. Different values of the hyperparameters were explored and those that allowed the best results to be obtained were selected. Finally, the model's measurement metrics were defined, in order to evaluate its accuracy and predictive capacity. The model designed in this article can provide a valuable tool for decision-making in the field of digital government and contribute to improving the efficiency and transparency in the performance of public officials.