Tradeoff and Sensitivity Analysis of a Hybrid Model for Ranking Commercial Off-the-Shelf Products

Abstract

Despite its popularity, The COTS-based development still faces some challenges, in particular the evaluation and selection process in which uncertainty plays a major role. A hybrid model, composed of the analytic hierarchy process (AHP) and Bayesian Belief Network (BBN), is proposed to evaluate and rank various COTS candidates while explicitly considering uncertainty. Several input parameters such as weights assigned to evaluation criteria, relative scores for various COTS candidates, and prior belief about the satisfaction of various attributes associated with the evaluation criteria need to be estimated. The estimation process of these input parameters is subject to uncertainty that limits the applicability of the model’s results. In this paper, we apply sensitivity analysis to check the validity and robustness of the model. Further, we apply tradeoff analysis to explore the impact of relaxing one criterion in order to achieve an increase in another criterion that is considered as more desirable in a particular project context. A digital library system is used as a case study to illustrate how the proposed tradeoff and sensitivity analysis was performed.