Corruption risk prediction in public procurement processes available on Datos Abiertos platform using Gradient Boosting Machine model

Abstract

Corruption and the mechanisms to combat it have become an issue of high importance for governments around the world, Colombia along with other OECD countries has open databases and public procurement systems that are constantly updated, and which are made up of a large amount of data, situation that represent a challenge for the control agencies faced with the task of auditing and analyzing such information. Machine Learning is then presented as a tool to facilitate the audit of public procurement, by offering a quick mechanism to identify those contractual processes with a higher risk of corruption. Thus, this article aims to evaluate the performance of a Gradient Boosting Machine model when predicting the risk of corruption in public procurement processes available on the Datos Abiertos platform.