Clasificación Automática del Avance de Megaproyectos de Inversión Pública en Colombia, desde un Enfoque Técnico, Organizacional y Ambiental

Abstract

Context: The TOE (Technical, Organizational, and Environmental) framework for the analysis of large scale projects is considered as the basis for the development of megaproject progress classification in accordance with the needs of the national planning agency in Colombia.Method: Classification of a megaproject progress is supported in the selection of several features taken from the TOE. These feature set is used to configure a database from the projects registered in the project-surveillance platform of the national planning agency in Colombia. The database is used to train two classification models. Information about 3200 projects from 2008 to 2012 was used, covering four economic sectors (Environment and sustainable development, Energy and mining, Health and social care and transportation). Debugging of the database was carried out by an analytic and quantitative approach. Model training and validation were computed with 70 % and 30 % of data respectively.Results: Obtained models have similar performances beyond 70 % in precision and agree in relevant input features.Conclusions: This work is a starting point to develop an automatic tool that can be used by the national planning agency of Colombia in the a-priori evaluation of delays in public investment Megaprojects.