Revisión y automatización de las planillas de un contrato de obra mediante técnicas de machine learning

Abstract

CONDITION FOR PUBLICATION OF PROJECT. integrative project focuses on the problem of the manual review of construction payrolls at the Fundación Terminal Terrestre de Guayaquil (FTTG), a process that entails operational inefficiencies and significant financial risks. Based on a risk management analysis and audit testing, the project seeks to identify opportunities for improvement and to develop a digital tool that, through the use of Machine Learning, ensures efficiency and transparency in the management of construction contracts. The methodology applied was structured into four stages: situational analysis, internal control evaluation, impact analysis, and improvement proposal. For the development of the project, relevant data from the contract and payrolls of the Municipal Costa Bus Terminal were utilized, complemented by interviews with key stakeholders involved in the payroll review process at FTTG. As part of the improvement proposal, a prototype digital platform was designed to automate the review process through the application of Machine Learning. The platform integrates several modules, including: file uploading in PDF and Excel formats, data processing, inconsistency detection through a predictive model, and automated report generation in Word format. The findings conclude that implementing an automated system is both feasible and necessary for FTTG. The proposed platform not only enhances efficiency and transparency in contract management but also strengthens the oversight of public resources by providing a reliable and agile tool that significantly improves current review processes. Keywords: Machine Learning, Construction Payrolls, COSO ERM 2017, Automation, Risk Management