作者: Andrea Saltelli , Paola Annoni
DOI: 10.1016/J.ENVSOFT.2010.04.012
关键词:
摘要: Mathematical modelers from different disciplines and regulatory agencies worldwide agree on the importance of a careful sensitivity analysis (SA) model-based inference. The most popular SA practice seen in literature is that 'one-factor-at-a-time' (OAT). This consists analyzing effect varying one model input factor at time while keeping all other fixed. While shortcomings OAT are known statistical literature, its widespread use among raises concern quality associated analyses. present paper introduces novel geometric proof inefficiency OAT, with purpose providing modeling community convincing possibly definitive argument against OAT. Alternatives to indicated which based theory, drawing experimental design, regression proper.