Three aspects on complex performance analysis under uncertainty

(eng) Data envelopment analysis (DEA) is a widely used non-parametric method for estimating the relative input-output efficiency for a set of homogenous decision making units (DMUs). Due to its solid underlying mathematical basis and wide applications to real-world problems, much effort has been devoted to DEA models. Though DEA as a well-known methodology provides many advantages relative to other approaches, there are some limitations, complexities and challenges that need to be addressed. It is therefore important to extend the DEA models to fit the real characteristics involving (i) the data generating process (DGP), (ii) the production process, and (iii) the evaluation need. Conventional DEA methods require accurate measurement of both input and output data. However, the observed values of the inputs and outputs in real-world problems are sometimes ambiguous, uncertain and imprecise. In the first aspect of this thesis, we deal with such imperfect data in measuring the managerial and operational efficiency of firms along with providing a robust ranking order when both input and output data are imprecise. In the second aspect, we show how to apply DEA methods in multilevel structures such as supply chains, also in the presence of uncertainty. In the third aspect, we develop a fuzzy DEA model with imprecise and ambiguous data in order to evolve the scope of application to a larger set of real-life problems. Finally, we propose a two-stage algorithm to extend the DEA model using a common set of input and output weights (CSW).