Bibliographic citations
Rojas, P., Acevedo, M. (2023). Propuesta de diseño de un modelo de machine learning para estimar costos en proyectos de software [Trabajo de investigación, Universidad Peruana de Ciencias Aplicadas (UPC)]. http://hdl.handle.net/10757/672209
Rojas, P., Acevedo, M. Propuesta de diseño de un modelo de machine learning para estimar costos en proyectos de software [Trabajo de investigación]. PE: Universidad Peruana de Ciencias Aplicadas (UPC); 2023. http://hdl.handle.net/10757/672209
@mastersthesis{renati/1297474,
title = "Propuesta de diseño de un modelo de machine learning para estimar costos en proyectos de software",
author = "Acevedo Mortola, Miguel Alberto",
publisher = "Universidad Peruana de Ciencias Aplicadas (UPC)",
year = "2023"
}
This thesis proposes the use of Business Analytics and Predictive Analytics techniques to solve the current problems of the company Gimalca S.A. regarding the cost estimation of projects led by project managers, which have presented problems in the scope, estimation and other reasons, this has led to losses of S/ 1.25MM, of which S/ 0.45MM correspond to errors in the estimation, which is not acceptable by the company. Likewise, as part of the research work, the use of the CRISP-DM methodology is proposed for the implementation of the analytical solution oriented to estimate the costs of the different projects evaluated by the company, the use of the methodology is recommended due to the integration between the business objectives and data mining. The solution proposes the development of predictive models to solve this problem, for which 4 models have been worked with the Google Colab tool, which uses the Python programming language, applying algorithms of Multiple Linear Regression, SVR (Support Vector Regression), Neural Networks and Bayesian Networks, being the Neural Networks algorithm the one with the best results. Therefore, it has been proved that through predictive models it is possible to predict the real cost associated to a project, which will allow to generate value for the company. Likewise, the thesis proposes the necessary guidelines for Gimalca S.A. to successfully apply the CRISP-DM methodology.
This item is licensed under a Creative Commons License