Modelo para seleção de projetos em gestão de portfólio de inovação utilizando analytic network process e um algoritmo de aprendizagem de máquina com regressão logística

Abstract

The decision making process for project selection and prioritization is considered to be the most relevant part of project portfolio management, and encompasses the project life cycle and the entire portfolio information flow, from idea evaluation to new product development. The selection and prioritization of portfolio management projects includes predetermined steps that involve the analysis of financial and non-financial criteria. This research performed a systematic review on the use of decision support methods in this process, considering 180 articles published in the period from 2009 to 2019. Results highlight which decision support methods are most often applied in project portfolio management and, in addition, some factors that led to the predominance of multi-criteria decision support methods, e.g. analytical hierarchy process / analytical network process. This paper is part of research on innovation portfolio management and contributes to scientific knowledge on this topic with an overview of the different decision support methods, an understanding of its operation, the most used methods in the last 10 years and the possible factors that impact on the school the method most used in project portfolio management, due to the results found in different articles. Furthermore, supports a computational model proposal using the predominant multicriteria methods.