Banca de DEFESA: MARCOS LUIZ LINS FILHO

Uma banca de DEFESA de DOUTORADO foi cadastrada pelo programa.
DISCENTE : MARCOS LUIZ LINS FILHO
DATA : 29/07/2019
HORA: 14:30
LOCAL: NEPSA II
TÍTULO:

MMCP-PUB/BR: MEASUREMENT MODEL OF PROJECTS COMPLEXITY FOR BRAZILIAN PUBLIC SECTOR

PALAVRAS-CHAVES:

 Project management. Project complexity. Measurement model. Machine Learning. Support Vector Machine.

PÁGINAS: 270
GRANDE ÁREA: Ciências Sociais Aplicadas
ÁREA: Administração
RESUMO:

Understanding the "Project Complexity" construct has challenged researchers and professionals since the study of Baccarini (1996) initiated in-depth discussions on the subject.  Within this context, the development of methods and tools, which can directly contribute to the insertion of the complexity dimension in the project management practices, has run into the difficulty of measuring it. In this sense, this research had as general objective to construct a measurement model of the Project Complexity for the Brazilian public sector. To do so, it was considered as the basis the perception of 16 specialists and 118 professionals in the area of management project who act or have already acted in public projects. The research was divided into two stages. In the first, we initially mapped a set of 96 complexity factors identified from a systematic review of the literature. These factors were evaluated through questionnaires by 16 experts and 118 professionals who identified the ten most relevant factors for the complexity of public projects. In the second stage, machine learning techniques were applied with support vector machines to construct a measurement model of the Project Complexity. In the model, three levels were defined to classify the complexity of a project: low, medium and high. The model was validated by 17 experiments divided in three rounds and was based on a set containing 36 test cases, where the average accuracy level of classification was evaluated. As for the research results, two new concepts were proposed, one for project management and the other for project complexity. It was proposed a theoretical model containing 11 dimensions and 96 project complexity factors based on a systematic review of the literature. It was identified the ten most relevant complexity factors for public projects in the opinion of Brazilian specialists and professionals in the area of project management. Finally, the MMCP-PUB / BR (Measurement Model of Projects Complexity for Brazilian Public Sector) was constructed and validated through machine learning techniques and refinements that resulted in 100% accuracy in the predictions made for 36 elements present in the test set. Based on the results, it was concluded that the use of machine learning to measure the complexity of projects demonstrated the feasibility and ability to contribute to increase the insertion of the dimension complexity in the project management practices.

MEMBROS DA BANCA:
Interno - 2668551 - ANDRE MORAIS GURGEL
Externo à Instituição - BENNY KRAMER COSTA - UNINOVE
Presidente - 1149367 - MANOEL VERAS DE SOUSA NETO
Externo ao Programa - 2575537 - MARCOS FERNANDO MACHADO DE MEDEIROS
Externo à Instituição - NICOLAU REINHARD - USP