An application of data envelopment analysis in telephone offices evaluation with partial data

Abstract The presence of partial data motivates the need to investigate how such factors can be incorporated into the existing measurement models. In this paper, a procedure is proposed for incorporating a set of factors with partial data into the DEA structure and restricting factor weights. The first DEA formulation is a complicated non-linear model issued from the set of partial data. In order to transform the first formulation into an ordinary linear programming model, both a linear scale transformation and variable change technique are used. The resulting linear programming model is then applied to the efficiency evaluation of telephone offices. Scope and purpose A procedure for handling both linear partial data on factor values and its weight preferences in data envelopment analysis (DEA) structure is presented. DEA is a methodology for driving the relative efficiencies of organizations or decision making units (DMUs) where there are multiple incommensurate inputs and outputs. The usual setting for many DEA applications involves a set of similar DMUs, for each of which there is an observable and measurable set of inputs and outputs. In some applications, however, a number of factors may be measurable only on partial data such as ordinal rankings and ratio bounds, owing mainly to intangible attributes to reflect social and environmental impacts. A model dealing with these partial data is presented and applied to the efficiency of telephone offices.