Data envelopment analysis for decision support

Abstract

PurposeTo provide decision makers (DMs) an option for addressing problems involving finite alternative sets and multiple criteria, where criterion weighting is difficult or impossible.Design/methodology/approachThe multicriteria decision problem is described, and a typically descriptive (rather than prescriptive) tool, data envelopment analysis (DEA), is summarized, along with a hypothetical but typical example of a multicriteria decision (vendor selection). The DEA approach is modified to incorporate weight constraints and is used to rank the available vendors. Results are compared with those from the use of a popular multicriteria decision tool (SMART) and a naïve averaging approach.FindingsThe modified DEA approach yields results very similar to those produced using SMART; these results are quite satisfactory in spite of the fact that DEA requires less involvement on the part of the DM. In addition, non‐dominant optima (a possible anomaly with DEA) are avoided, and often a single alternative, rather than a non‐dominated set, will result, thus providing a unique optimum.Research limitations/implicationsResults are based on the analysis of a single data set. Future investigation should examine the performance of the DEA approach when other data sets involving more like as well as more unlike alternatives are involved.Practical implicationsWith DEA the burden on the DM is reduced, as the need for eliciting criterion weights is obviated. DEA should thus provide an acceptable alternative to prescriptive modeling tools when multiple DMs are involved and/or criterion weight determination is unfeasible.Originality/valueThis paper demonstrates how DEA, a tool used more typically in post hoc evaluations, can be used also, with some modifications, as a prescriptive decision support tool.