portfolio decision analysis; multicriteria decision making/aiding; FITradeoff; partial information; metaheuristics; project portfolio problem.
The current global scenario, marked by resource scarcity, uncertainties, and challenges in managing multiple criteria, is often associated with the complexity of project portfolio selection. The efficient generation of portfolios that optimize resource usage and achieve strategic objectives is a complex task. This work presents a new proposal for the flexible and interactive FITradeoff method in portfolio decision analysis, incorporating partial information about decision-makers' preferences. In this context, two approaches were proposed. The first involves the explicit generation of portfolios, while the second utilizes metaheuristics to optimize the portfolio generation process. The first proposed approach specifically uses the concept of c-optimal portfolios and refinement strategies of feasibility and efficiency during the process of generating a portfolio while endeavoring to keep both computational and cognitive efforts within reasonable limits. This thesis also proposes novel metaheuristic-based multiattribute decision approach reduce the solution space when generating feasible and efficient project portfolios incorporating the decision-maker’s preferences. The FITradeoff approach to combinatorial portfolio is incorporated into a Decision Support System (DSS). The ability to operate via the DSS, developed as an integral part of this thesis, stands out as a significant contribution, providing a practical implementation of the proposed methods. The results share relevant evidence, demonstrating that the computational outcomes using the proposed methods exhibit good performance in terms of minimizing computational effort and reducing the cognitive effort required of the decision-maker. While the two proposed approaches address the challenge of portfolio generation from distinct perspectives, they complement each other, offering insights that can be integrated into different contexts. They are flexible, allowing replication in diverse situations. This synergy between the approaches provides a comprehensive framework that can be effectively applied in practical situations, consolidating the utility and adaptability of these models for solving the project portfolio problem.