A ROBUST OPTIMIZATION MODEL FOR PROJECT PORTFOLIO SELECTION WITH INFORMATION UNCERTAINTY

For the project portfolio selection problem with information uncertainty, traditional methods such as deterministic programming and stochastic programming are not applicable. Under uncertain data, to pursue an absolutely optimal solution is impossible and hence robust solutions are more rational since they remain "close" to optimality (i.e., solution robust) and "almost" feasible (i.e., model robust) for all data scenarios. Therefore, a new modeling approach is needed to provide such robust recommendations. In this paper, a robust optimization mathematical programming model is developed for project portfolio selection with information uncertainty. Introducing the uncertain parameters (e.g., project's return, demand for a certain resource, and etc) into a general deterministic zero-one model, the robust optimization formulation problem is seriously addressed. Robust solutions can then be obtained and analyzed. The trade-off between the probability of solution's feasibility and the loss of optimality is also discussed. A case is used to illustrate the application of the proposed model.