Review of ranking methods in the data envelopment analysis context

Size: px
Start display at page:

Download "Review of ranking methods in the data envelopment analysis context"

Transcription

1 European Journal of Operational Research 140 (2002) Review of ranking methods in the data envelopment analysis context Nicole Adler a, *,1, Lea Friedman b,2, Zilla Sinuany-Stern b,c,2 a School of Business Administration, Hebrew University of Jerusalem, Mount Scopus, Jerusalem 91905, Israel b Department of Industrial Engineering and Management, Ben-Gurion University of the Negev, P.O. Box 653, Beer-Sheva 84105, Israel c The College of Judea and Samaria, Ariel, Israel Abstract Within data envelopment analysis (DEA) is a sub-group of papers in which many researchers have sought to improve the differential capabilities of DEA and to fully rank both efficient, as well as inefficient, decision-making units. The ranking methods have been divided in this paper into six, somewhat overlapping, areas. The first area involves the evaluation of a cross-efficiency matrix, in which the units are self and peer evaluated. The second idea, generally known as the super-efficiency method, ranks through the exclusion of the unit being scored from the dual linear program and an analysis of the change in the Pareto Frontier. The third grouping is based on benchmarking, in which a unit is highly ranked if it is chosen as a useful target for many other units. The fourth group utilizes multivariate statistical techniques, which are generally applied after the DEA dichotomic classification. The fifth research area ranks inefficient units through proportional measures of inefficiency. The last approach requires the collection of additional, preferential information from relevant decision-makers and combines multiple-criteria decision methodologies with the DEA approach. However, whilst each technique is useful in a specialist area, no one methodology can be prescribed here as the complete solution to the question of ranking. Ó 2002 Published by Elsevier Science B.V. Keywords: Data envelopment analysis; Ranking techniques; Super-efficiency; Cross-efficiency; Benchmarking; Multivariate statistics; MCDM 1. Introduction DEA was initiated in 1978 when Charnes, Cooper and Rhodes (CCR) demonstrated how to * Corresponding author. Tel.: address: msnic@mscc.huji.ac.il (N. Adler). 1 I would like to thank the Kreitman Foundation Post- Doctoral Fellowship at Ben-Gurion University of the Negev and the partial support of the Recanati Foundation. 2 This paper was partially supported by the Paul Ivanier Center for Robotics and Production Management, Ben-Gurion University of the Negev. change a fractional linear measure of efficiency into a linear programming (LP) format. As a result, decision-making units (DMUs) could be assessed on the basis of multiple inputs and outputs, even if the production function was unknown. Indeed, this paper will only consider the multiple inputs and multiple outputs case. This non-parametric approach solves an LP formulation per DMU and the weights assigned to each linear aggregation are the results of the corresponding LP. The weights are chosen so as to show the specific DMU in as positive a light as possible, /02/$ - see front matter Ó 2002 Published by Elsevier Science B.V. PII: S (02)

2 250 N. Adler et al. / European Journal of Operational Research 140 (2002) under the restriction that no other DMU, given the same weights, is more than 100% efficient. Consequently, a Pareto frontier is attained, marked by specific DMUs on the boundary envelope of input output variable space. The frontier is considered a sign of relative efficiency, which has been achieved by at least one DMU. Charnes et al. (1978) described DEA as a mathematical programming model applied to observational data [which] provides a new way of obtaining empirical estimates of extremal relations such as the production functions and/or efficient production possibility surfaces that are a cornerstone of modern economics. Various theoretical extensions have been developed, based on the original CCR model: Banker et al. (1984) developed a variable returnsto-scale variation; the multiplicative model was developed by Charnes et al. (1982) in which the data are transformed using a logarithmic structure; Charnes et al. (1985b) developed the additive variation, in which the objective function contains slack variables alone. Seiford and Thrall (1990) provide a useful discussion and comparison of all the basic models available to date in DEA. Many additional theoretical papers in the field have adapted the models to deal with problems that have occurred in practice. One adaptation has been in the field of ranking DMUs. The basic DEA results group the DMUs into two sets, those that are efficient and define the Pareto frontier and those that are inefficient. In order to rank all the DMUs, another approach or modification was required. Often decision-makers are interested in a complete ranking, beyond the dichotomized classification, in order to refine the evaluation of the units. One problem that has been discussed frequently in the literature has been the lack of discrimination in DEA applications, in particular when there are insufficient DMUs or the number of inputs and outputs is too high relative to the number of units. This is an additional reason for the growing interest in complete ranking techniques. Furthermore, fully ranking units is an established approach in the social sciences, in general (see Young and Hamer, 1987), and in multiple-criteria decision making (MCDM), in particular. By now many papers on DEA and ranking (over 50) have been published over the last decade within the DEA context. Since most decision-makers are interested in a complete ranking, the use of such techniques aids in marketing the DEA approach. It should be noted that the methods discussed here could be considered postanalyses since they do not replace the standard DEA models but rather provide added value. This review describes the ranking methods developed in the literature and since many papers have been published in this field, we have grouped them into six basic areas. These methods are classified by several criteria and are not entirely mutually exclusive. After specifying the DEA method mathematically in Section 2, Section 3 discusses the cross-efficiency technique, which was first suggested by Sexton et al. (1986), whereby the DMUs are both self and peer evaluated. In Section 4 we will review the super-efficiency technique, first published in Andersen and Petersen s paper of 1993, in which DMUs are ranked through the exclusion of the unit being scored from the DEA dual LP. Section 5 discusses the evaluation of DMUs through benchmarking, an approach originating in Torgersen et al. (1996), in which an efficient unit is highly ranked if it appears frequently in the reference sets of inefficient DMUs. Section 6 will review the papers that apply multivariate statistical tools, such as canonical correlation analysis and discriminant analysis, in order to reach a complete DMU rank, first suggested by Friedman and Sinuany-Stern (1997). Section 7 discusses the ranking of inefficient DMUs, not always considered in the previous methodologies. Section 8 discusses the cross-pollination between multi-criteria decision-making (MCDM) methodologies and DEA in terms of ranking. Finally, Section 9 presents the results of the various methodologies on an example discussed in Sexton et al. (1986) and Section 10 provides conclusions and a summary of the review, including a table of the DEA software packages available today and the ranking approaches they offer automatically. 2. The data envelopment analysis model DEA is a mathematical model that measures the relative efficiency of decision-making units

3 N. Adler et al. / European Journal of Operational Research 140 (2002) with multiple inputs and outputs but with no obvious production function to aggregate the data in its entirety. Relative efficiency is defined as the ratio of total weighted output to total weighted input. By comparing n units with s outputs denoted by y rk ; r ¼ 1;...; s; and m inputs denoted by x ik, i ¼ 1;...; m; the efficiency measure for DMU k is h k ¼ Max u r;v i P s u ry rk P m v ix ik ; where the weights, u r and v i, are non-negative. A second set of constraints requires that the same weights, when applied to all DMUs, do not provide any unit with efficiency greater than one. This condition appears in the following set of constraints: P s u ry rj P m v ix ij 6 1 for j ¼ 1;...; n: The efficiency ratio ranges from zero to one, with DMU k being considered relatively efficient if it receives a score of one. Thus, each unit will choose weights so as to maximize self-efficiency, given the constraints. The result of the DEA is the determination of the hyperplanes that define an envelope surface or Pareto frontier. DMUs that lie on the surface determine the envelope and are deemed efficient, whilst those that do not are deemed inefficient. The formulation described above can be translated into a linear program, which can be solved relatively easily and a complete DEA solves n linear programs, one for each DMU. h k ¼ Max X m X m v i x ij Xs X s v i x ik ¼ 1; u r y rk u r P 0 for r ¼ 1;...; s; v i P 0 for i ¼ 1;...; m: u r y rj P 0 for j ¼ 1;...; n; ð1þ Model (1), often referred to as the CCR model (Charnes et al., 1978), assumes that the production function exhibits constant returns-to-scale. The BCC (Banker et al., 1984) model adds an additional constant variable, c k, in order to permit variable returns-to-scale: h k ¼ Max X m X m v i x ij Xs Xs v i x ik ¼ 1; u r y rk þ c k u r P 0 for r ¼ 1;...; s; v i P 0 for i ¼ 1;...; m: u r y rj c k P 0 for j ¼ 1;...; n; ð2þ It should be noted that the results of the CCR input-minimized or output-maximized formulations are the same, which is not the case in the BCC model. Thus, in the output-oriented BCC model, the formulation maximizes the outputs given the inputs and vice versa The dual program of the CCR model If a DMU proves to be inefficient, a combination of other efficient units can produce either greater output for the same composite of inputs, use fewer inputs to produce the same composite of outputs or some combination of the two. A hypothetical decision making unit, k 0, can be composed as an aggregate of the efficient units, referred to as the efficient reference set for inefficient unit k. The solution to the dual problem of the linear program directly computes the multipliers required to compile k 0 : Min f k Xn X n L kj x ij þ f k x ik P 0 for i ¼ 1;...; m; L kj y rj P y rk for r ¼ 1;...; s; L kj P 0 for j ¼ 1;...; n: ð3þ

4 252 N. Adler et al. / European Journal of Operational Research 140 (2002) In the case of an efficient DMU, all dual variables will equal zero except for L kk and f k, which reflect the unit k s efficiency, both of which will equal one. If DMU k is inefficient, f k will equal the ratio solution of the primal problem. The remaining variables, L kj, if positive, represent the multiples by which unit k s inputs and outputs should be multiplied in order to compute the composite efficient DMU k The slack-adjusted CCR model In the slack-adjusted DEA models, see for example model (4), a weakly efficient DMU will now be evaluated as inefficient, due to the presence of input and output oriented slacks s i and r r, respectively.! X m Min f k e s i þ Xs r r Xn X n L kj x ij þ f k x ik s i ¼ 0 for i ¼ 1;...; m; L kj y rj r r ¼ y rk for r ¼ 1;...; s; L kj ; s i ; r r P 0 for j ¼ 1;...; n; ð4þ whereby e is a positive non-archimedean infinitesimal The additive model An alternative formulation proposed by Charnes et al. (1985b) utilizes slacks alone in the objective function. This model is used in both the benchmarking approach and measure of inefficiency dominance developed in Sections 5 and 7, respectively. Min Xm Xn X n s i Xs r r L kj x ij s i ¼ x ik for i ¼ 1;...; m; L kj y rj r r ¼ y rk for r ¼ 1;...; s; L kj ; s i ; r r P 0 for j ¼ 1;...; n: ð5þ In order to avoid large variability in the weights for all DEA models, Thompson et al. (1986, 1990, 1992) suggest the use of assurance regions (ARs) i.e. bounds on the weights. This, in turn, increases the differentiability among the unit scores by reducing the number of efficient DMUs. In the extreme case, the weights will be reduced to a single set of common weights and the units will be fully ranked. However, the AR literature is not discussed in this review since the concept does not strive, nor does it generally succeed, in reaching a complete ranking of DMUs. Another attempt to improve the discriminating power of DEA can be found in Adler and Golany (2001), where principal component analysis was utilized to reduce the number of inputs/outputs, thus reducing the problem of dimensionality. However, this technique cannot ensure a complete ranking but rather a reduction in the set of efficient units. 3. Cross-efficiency ranking methods The cross-evaluation matrix was first developed by Sexton et al. (1986), inaugurating the subject of ranking in DEA. Indeed, as Doyle and Green (1994) argued, decision-makers do not always have a reasonable mechanism from which to choose assurance regions, thus they recommend the crossevaluation matrix for ranking units. The crossefficiency method simply calculates the efficiency score of each DMU n times, using the optimal weights evaluated by the n LPs. The results of all the DEA cross-efficiency scores can be summarized in a cross-efficiency matrix as shown in Eq. (6): P s h kj ¼ P u rky rj m v ; k ¼ 1;...; n; j ¼ 1;...; n: ikx ij ð6þ Thus, h kj represents the score given to unit j in the DEA run of unit k i.e. unit j is evaluated by the weights of unit k. Note that all the elements in the matrix are between zero and one, 0 6 h kj 6 1, and the elements in the diagonal, h kk, represent the standard DEA efficiency score, h kk ¼ 1 for efficient units and h kk < 1 for inefficient units. Further-

5 N. Adler et al. / European Journal of Operational Research 140 (2002) more, if the weights of the LP are not unique, a goal-programming technique can be applied to choose between the optimal solutions. According to Sexton et al. (1986), the secondary goals could be, for example, either aggressive or benevolent. In the aggressive context, DMU k chooses amongst the optimal solutions such that it maximizes selfefficiency and at a secondary level minimizes the other DMUs cross-efficiency levels. The benevolent secondary objective would be to maximize all DMUs cross-efficiency rankings. See also Oral et al. (1991) for methods of evaluating the goal programs. The cross-efficiency ranking method in the DEA context utilizes the results of the cross-efficiency matrix h kj in order to rank scale the units. Let us define h k ¼ P n h kj=n as the average crossefficiency score given to unit k. Averaging, however, is not the only possibility, as the median, minimum or variance of scores could also be applied (see Green et al. (1996) for further detailed suggestions to help in avoiding such problems as the lack of Independence of Irrelevant Alternatives). It could be argued that h k,oran equivalent, is more representative than h kk, the standard DEA efficiency score, since all the elements of the cross-efficiency matrix are considered, including the diagonal. Furthermore, all the units are evaluated with the same sets of weight vectors. Thus the h k score better represents the unit evaluation since it measures the overall ratios over all the runs of all the units. The maximum value of h k is 1, which occurs if unit k is efficient in all the runs i.e. all the units evaluate unit k as efficient. In order to rank the units, we can simply assign the unit with the highest score a rank of one and the unit with the lowest score a rank of n. While the DEA scores, h kk, are non-comparable, since each uses different weights, the h k score is comparable because it utilizes the weights of all the units equally. However, this is also the drawback of the technique, since the evaluation subsequently loses its connection to the multiplier weights. Furthermore, Doyle and Green (1994) developed the idea of a maverick index within the confines of cross-efficiency. The index measures the deviation between h kk, the self-appraised score and the unit s peer scores, as shown in Eq. (7). M k ¼ h kk e k ; where e k ¼ 1 X h kj : ð7þ e k ðn 1Þ The higher the value of M k, the more the DMU can be considered a maverick. Doyle and Green (1994) argued that this score can go handin-hand with the benchmarking process (Section 5), whereby those DMUs considered efficient under self-appraisal but fail to appear in the reference sets of inefficient DMUs will generally achieve a high M k score. Those achieving a low score are generally all-round performers and are frequently both self- and peer efficient. 4. Super-efficiency ranking techniques Andersen and Petersen (1993) developed a new procedure for ranking efficient units. The methodology enables an extreme efficient unit k to achieve an efficiency score greater than one by removing the kth constraint in the primal formulation, as shown in model (8). h k ¼ Max X m X m v i x ij Xs Xs v i x ik ¼ 1; u r y rk u r Pe for r ¼ 1;...;s; v i Pe for i ¼ 1;...;m: j6¼k u r y rj P0 for j ¼ 1;...;n; j 6¼ k; ð8þ The dual formulation of the super-efficient model, as seen in model (9), computes the distance between the Pareto frontier, evaluated without unit k, and the unit itself i.e. for J ¼ fj ¼ 1;...; n; j 6¼ kg.

6 254 N. Adler et al. / European Journal of Operational Research 140 (2002) Min f k X L kj x ij 6 f k x ik for i ¼ 1;...; m; j2j X L kj y rj P y rk for r ¼ 1;...; s; j2j L kj P 0 for j ¼ 1;...; n: ð9þ However, there are three problematic areas with this methodology. First, Andersen and Petersen refer to the DEA objective function value as a rank score for all units, despite the fact that each unit is evaluated according to different weights. This value in fact explains the proportion of the maximum efficiency score that each unit k attained with its chosen weights in relation to a virtual unit closest to it on the frontier. Furthermore, if we assume that the weights reflect prices, then each unit has different prices for the same set of inputs and outputs within the same organization. Second, the super-efficient methodology can give specialized DMUs an excessively high ranking. To avoid this problem, Sueyoshi (1999) introduced specific bounds on the weights in a super-efficient ranking model as described in Eqs. (10). v i P 1=ðm þ sþ max ðx ij Þ j ð10þ u r P 1=ðm þ sþ max ðy rj Þ: j Furthermore, in order to limit the super-efficient scores to a scale with a maximum of 2, Sueyoshi developed an Adjusted Index Number (AIN) formulation, as shown in Eq. (11). AIN k ¼ 1 þ d k min j2e d j max j2e d j where E is the set of efficient units: min j2e d j ; ð11þ The third problem lies with an infeasibility issue, which if it occurs, means that the super-efficient technique cannot provide a complete ranking of all DMUs. Thrall (1996) used the model to identify extreme efficient DMUs and noted that the superefficiency CCR model may be infeasible. Zhu (1996a), Dula and Hickman (1997) and Seiford and Zhu (1999) prove under which conditions various super-efficient DEA models are infeasible. Mehrabian et al. (1999) suggested a modification to the dual formulation in order to ensure feasibility, as specified in model (12). Min f k X L kj x ij þ x ik f k P 0 for i ¼ 1;...; m; j2j X L kj y rj P y rk for r ¼ 1;...; s; j2j L kj P 0 for j ¼ 1;...; n: ð12þ Despite these drawbacks, possibly because of the simplicity of the concept, many published papers have used this approach. For example, Hashimoto (1997) developed a DEA super-efficient model with assurance regions in order to rank the DMUs completely. Using model (13), Hashimoto avoided the need for compiling additional preference information in providing a complete ranking of n candidates. h k ¼ Max X s X s u r y rk u r y rj 6 1 for j ¼ 1;...; n; j 6¼ k; u r u rþ1 P e for r ¼ 1;...; s 1; u s P e; u r 2u rþ1 þ u rþ2 P 0 for r ¼ 1;...; s 2; ð13þ where u r is the sequence of weights given to the rth place vote (whereby each voter selects and ranks the top t candidates). The use of assurance regions avoids the specialization pitfall of the standard super-efficiency model. 5. Benchmark ranking method Torgersen et al. (1996) achieved a complete ranking of efficient DMUs by measuring their importance as a benchmark for inefficient DMUs.

7 N. Adler et al. / European Journal of Operational Research 140 (2002) The benchmarking measure is evaluated in a twostage procedure, whereby the additive model (see Section 2.3) is first used to evaluate the value of the slacks. The set of efficient units, V, is identified as those units whose slack values equal zero. In the second stage, model (14) is applied to all decisionmaking units. 1 E k ¼ Max f k X L kj x ij s ik ¼ x ik for i ¼ 1;...; m; j2v X L kj y rj f k y rk r rk ¼ 0 for r ¼ 1;...; s; j2v X L kj ¼ 1: j2v ð14þ In order to rank the efficient DMUs and evaluate which are of particular importance to the industry, the benchmarking measure aggregates the individual reference weights as shown in Eq. (15). P n q r k L jk yrj P y rj yr P y r 8k ¼ 1;...; V ; r ¼ 1;...; s; where ð15þ y P rj ¼ y rj þ r rj : E j For each efficient DMU k, the benchmark q r k measures the fraction of total aggregated potential increase in output r, over which k acts as a referent. The efficient units together define the entire potential within each output variable. An average value of q k can then be calculated in order to rank all efficient DMUs completely. Torgersen et al. (1996) apply their technique to a set of unemployment offices in Norway and show that in certain cases different conclusions were reached to that of Andersen and Petersen s super-efficiency method (see Section 4), due to the outlier problem of the latter technique. This is somewhat similar to the results reported in Sinuany-Stern et al. (1994), in which an efficient DMU is highly ranked if it is chosen as a useful target by many other inefficient units. The technique developed in this paper is applied to all DMUs in two stages. In the first stage, the efficient units are ranked by simply counting the number of times they appear in the reference sets of inefficient units, an idea first developed in Charnes et al. (1985a). The inefficient units are then ranked, in the second stage, by counting the number of DMUs that needs to be removed from the analysis before they are considered efficient. However, a complete ranking cannot be assured since many DMUs may receive the same ranked score. 6. Ranking with multivariate statistics in the DEA context An alternative approach suggested in the literature involves the use of statistical techniques in alliance with DEA to achieve a complete ranking. One of the major aims of the methodologies described in this section is to close the gap between DEA and the classical statistical approaches. It should be noted that DEA is a methodology directed towards frontiers rather than central tendencies. Instead of trying to fit regression planes through the center of the data, DEA floats a piecewise linear surface (the efficient frontier) that rests on top of the observations. DEA optimizes and thereby focuses on each unit separately, while regression is a parametric approach that fits a single function to the data collected on the basis of average behavior that requires the functional form to be pre-specified. On the other hand, in DEA the values of the weights differ from unit to unit and whilst this flexibility in the choice of weights characterizes the DEA model, different weights cannot generally be used for ranking because these scores are obtained from different comparison (peer) groups for different units. In this section we present three ranking procedures based on multivariate statistical analyses using common weights. Non-parametric statistical tests, such as Mann Whitney, are utilized throughout to verify the compatibility between the rank and the DEA dichotomic classification. Empirically, the literature shows a high statistical

8 256 N. Adler et al. / European Journal of Operational Research 140 (2002) significance between the statistical analyses and the DEA results Canonical correlation analysis for ranking Canonical correlation is an extension of regression analysis. While the regression model explains a single output using multiple inputs, canonical correlation analyzes multiple inputs and multiple outputs. CCA searches for a single vector weight for the inputs and outputs, common to all the units. CCA constructs a composite input variable Z j, as a linear combination of the m inputs and a composite output variable W j, as a linear combination of the s outputs: Z j ¼ V 1 x 1j þ V 2 x 2j þþv m x mj ; W j ¼ U 1 y 1j þ U 2 y 2j þþu s y sj ; where CCA determines the two vectors of coefficients, V 0 ¼ðV 1 ; V 2 ;...; V m Þ and U 0 ¼ðU 1 ; U 2 ;...; U s Þ, so as to maximize r zw, the coefficient of correlation between the composite input, Z, and the composite output, W. Formally, this is specified in model (16). Max V 0 S xx V ¼ 1; U 0 S yy U ¼ 1; V 0 S xy U r zw ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðv 0 S xx V ÞðU 0 S yy UÞ ð16þ where S xx ; S yy and S xy are the matrices of the sums of squares and sums of products of the variables, respectively. Note that the weights V 0 and U 0 are determined up to a proportional constant. There is a closed form solution for the weights, however, this ranking is only feasible in the DEA context if all the weights are non-negative. Friedman and Sinuany-Stern (1997) use the CCA method methodology by defining a scaling ratio score, T, as a ratio of linear combinations of the inputs and outputs. Then they utilize the common weights for the linear combinations that are drawn from the largest eigenvalue of the CCA method, as shown in Eq. (17). T j ¼ W P s j ¼ P U ry rj m Z j V ; j ¼ 1;...; n: ð17þ ix ij Note that while the DEA efficiency ratio h j is bounded above by 1, the scaling ratio T j of the CCA/DEA is unbounded. Thus, it is the rank order of the scaling ratios that is important rather than their absolute value. Although the CCA is independent of the DEA results, the same formulation has been applied to the two types of variables; weighted inputs and weighted outputs. Moreover, the scaling ratio score T is also similar to the DEA score, namely it represents the ratio described in DEA. The use of a single set of weights is possibly the most obvious methodology for ranking and in this case the CCA has been built in the light of the DEA context. Hence, empirically the results are statistically significantly similar Linear discriminant analysis for ranking Sinuany-Stern et al. (1994) used linear discriminant analysis (DDEA) in order to find a score function which ranks DMUs, given the DEA division into efficient and inefficient sets. Using traditional discriminant analysis of two groups in the DEA context, they constructed a one-dimensional linear function, as specified in Eq. (18). D j ¼ Xs u r y rj þ Xm v i ð x ij Þ; j ¼ 1;...; n: ð18þ If D j > D c, then unit j is classified as efficient, otherwise it is classified as inefficient, where D c is a critical value based on the midpoint of the means of the discriminant function value of the two groups (see Morrison, 1976). The value of D j is used for ranking, where the unit with the largest value is the most efficient etc. The common weights u r and v i have closed form solutions, however, the DDEA ranking is feasible only if all the weights are non-negative. Furthermore, as with the CCA/DEA formulation, the results of the DDEA will only be statistically similar to the standard DEA result, which means that some efficient units may be ranked lower than inefficient units and vice versa.

9 N. Adler et al. / European Journal of Operational Research 140 (2002) Discriminant analysis of ratios for ranking In addition, Sinuany-Stern and Friedman (1998) developed another technique in which discriminant analysis of ratios was applied to DEA (DR/DEA) in order to avoid the infeasibility problems of the previous two approaches. Instead of considering a linear combination of the inputs and outputs in one equation (as in the traditional discriminant analysis of the two groups), they constructed a ratio function between a linear combination of the inputs and a linear combination of the outputs. In some ways this ratio function is similar to the DEA efficiency ratio, however, whilst DEA provides weights for the inputs and outputs, which vary from unit to unit, DR/DEA provides common weights for all units. In principle, DR/DEA determines the weights such that the ratio score function discriminates optimally between two groups of observations (DMUs) on a one-dimensional scale (in our case, efficient and inefficient units predetermined by DEA). The ratio, T j, and the arithmetic means of the ratio scores of the efficient and inefficient groups are P s T j ¼ P u ry rj m v ; j ¼ 1;...; n; ix ij T 1 ¼ Xn 1 T j n 1 and T 2 ¼ Xn j¼n 1 þ1 T j n 2 ; where n 1 and n 2 are the number of efficient and inefficient units in the DEA model, respectively. The weighted mean of the entire n units ðn ¼ n 1 þ n 2 Þ will be denoted by T ¼ n 1T 1 þ n 2 T 2 : n Our problem is to find the common weights v i and u r such that the ratio of the between-group variance of T ; ðss B ðt ÞÞ and the within group variance of T ; ðss W ðt ÞÞ will be maximized, as shown in model (19). SS B ðt Þ max k ¼ max u r;v i u r;v i SS W ðt Þ ; ð19þ SS B ðt Þ¼n 1 ðt 1 T Þ 2 þ n 2 ðt 2 T Þ 2 ¼ n 1n 2 n 1 þ n 2 ðt 1 T 2 Þ 2 ; SS W ðt Þ¼ Xn 1 ðt j T 1 Þ 2 þ Xn j¼n 1 þ1 ðt j T 2 Þ 2 : DR/DEA constructs the efficiency score for each unit j as T j, the ratio between the composite output and composite input. Thus it rank scales the DMUs so that the unit with the highest score receives rank 1 and the unit with the lowest score ranks n. If any weight is negative then non-negativity constraints ought to be added to the optimization problem. To solve this problem, they used a non-linear search optimization algorithm, however, there is no guarantee that the solution found is globally optimal. 7. The ranking of inefficient decision-making units The majority of techniques so far discussed have not attempted to rank the inefficient decision-making units beyond the efficiency scores attained from the standard DEA models. However, as argued in Cooper and Tone (1997), the original efficiency value will generally be determined from different facets, which means that these values are being derived from comparisons involving performances of different sets of DMUs. The benchmarking concept only attempts to rank the efficient DMUs identified in the standard DEA models. The super-efficiency method ranks inefficient units in the same manner as the standard DEA model. It should be noted that both the cross-efficiency method and the statistical techniques do attempt to address this problem. One concept, derived in Bardhan et al. (1996), attempts to rank inefficient units using a Measure of Inefficiency Dominance (MID). The measure is based on slack-adjusted DEA models from which an overall measure of inefficiency can be computed as shown in Eq. (20). P m ðs i =x ikþþ P s ðr r =y rkþ 6 1: ð20þ m þ s The MID index ranks the inefficient DMUs according to their average proportional inefficiency in all inputs and outputs. However, just as the

10 258 N. Adler et al. / European Journal of Operational Research 140 (2002) benchmarking approach (see Section 5) only ranks the efficient units, the MID index only ranks the inefficient units. 8. DEA and multi-criteria decision-making methods The MCDM literature was entirely separate from DEA research until 1988, when Golany combined interactive, multiple-objective linear programming and DEA. Whilst the MCDM literature does not consider a complete ranking as their ultimate aim, they do discuss the use of preference information to further refine the discriminatory power of the DEA models. In this manner, the decision-makers could specify which inputs and outputs should lend greater importance to the model solution. However, this could also be considered the weakness of these methods, since additional knowledge on the part of the decision-makers is required. Golany (1988), Kornbluth (1991), Thanassoulis and Dyson (1992), Golany and Roll (1994), Zhu (1996b) and Halme et al. (1999) each incorporated preferential information into the DEA models through, for example, a selection of preferred input/output targets or hypothetical DMUs. A separate set of papers reflected preferential information through limitations on the values of the weights (assurance regions or cone-ratio models), which can almost guarantee a complete DMU ranking. Such papers include Thompson et al. (1986), Dyson and Thanassoulis (1988), Charnes et al. (1990, 1989), Cook and Kress (1990a,b), Thompson et al. (1990), Wong and Beasley (1990), Cook and Johnston (1992) and Green and Doyle (1995). Cook and Kress (1990a,b, 1991, 1994) and Cook et al. (1993, 1996) argued that by imposing ratio constraints on the multipliers and replacing the infinitesimal with a lower bound thus acting as a discrimination factor, the modified DEA model can almost ensure a single efficient DMU. For example, when considering aggregation of votes whereby y rj is the number of rth placed votes received by candidate j, one can define a discrimination intensity function dðr; eþ and solve model (21). Max e X s u r y rj 6 1; j ¼ 1;...; n; u r u rþ1 dðr; eþ P 0; u r dðr; eþ P 0; u r ; e P 0; ð21þ where dðr; eþ ensures that first place votes are valued at least as highly as second place votes and so on. The ease with which this formulation can be solved depends on the form of dðr; eþ. Model (21) is linear if the difference between the ranks is linear, but this need not always be the case. However, as pointed out in Green et al. (1996), the form of dðr; eþ affects the ranking results and no longer allows DMUs to choose their weights freely. Furthermore, which mathematical function is appropriate is unclear yet important to the analysis. Two papers have discussed the use of fuzzy logic in conjunction with DEA, both to incorporate preference information from experts and to fully rank the efficient DMUs. Both Karsak (1998) and Hougaard (1999) separately developed an essentially two-stage methodology in which DEA identifies the efficiency of the DMUs concerned and the second stage utilizes fuzzy logic to rank the efficient units based on expert knowledge and the more objective DEA findings. Karsak discusses a robot selection procedure and argues that the model can be applied to other engineering problems such as selection of sites or flexible manufacturing systems. Some have gone as far as to argue that DEA should be considered another methodology within MCDM, for example Troutt (1995), Li and Reeves (1999) and Sinuany-Stern et al. (2000). Troutt (1995) developed a maximin efficiency ratio model in which a set of common weights is evaluated to distinguish between efficient DMUs as shown in model (22). Maximize u r;v i Minimize k P s u ry rk P m v ix ik

11 N. Adler et al. / European Journal of Operational Research 140 (2002) P s u ry rk P m v ix ik 6 1 8k; X s u r ¼ 1; u r ; v i P 0 8r; i: ð22þ Li and Reeves (1999) suggested utilizing multiple objectives, such as minimax and minisum efficiency in addition to the standard DEA objective function in order to increase discrimination between DMUs, as shown in the multiple-objective linear program (MOLP) of model (23). Min d k ; Min M; Min X n Xm X m d j ; v i x ij þ Xs v i x ik ¼ 1; u r y rj þ d j ¼ 0 for j ¼ 1;...; n; M d j P 0; u r ; v i ; d j 6 0 8r; i; j: ð23þ The first objective is equivalent to the standard CCR model. The second objective function requires the MOLP to minimize the maximum deviation (slack variable) and the third objective is to minimize the sum of deviations. The aim was to increase discrimination, which the second and third objectives provide without promising complete ranking, similarly to the assurance regions and cone-ratio approaches. On the other hand, this approach does not require additional preferential information as do other approaches. Joro et al. (1998) provide a review of the literature on DEA and MOLP. Sinuany-Stern et al. (2000) suggested an analytical hierarchical process (AHP/DEA) ranking model based on a two-stage process. In the first stage, a DEA model is run for every pair of DMUs, two units at a time, ignoring all others. Based on the results of the first stage, a pairwise comparison matrix is created from which a single level AHP can be applied thus providing a full scale ranking of all DMUs. Since the basic data are pre-specified (inputs and outputs of organizational units) the axiomatic utility theory problems of AHP do not exist. Indeed, in the original AHP, the pairwise comparison matrix data are based on the subjective decision-makers preferences whilst AHP/DEA builds an objective matrix. This nonsubjective approach is easier from the decisionmakers viewpoint since there is no subjective evaluation of many pairs of alternatives, another drawback of AHP. In the AHP/DEA model, the multiple criteria are taken into account through DEA while the ranking is performed by AHP, thus the model does not suffer from the limitations of either model. However, AHP can result in a matrix of many 1 s, which would result in a failure to fully rank the DMUs. However, it should be noted that certain researchers have argued that MCDM and DEA are two entirely separate approaches, which do not overlap. MCDM is generally applied to ex ante problem areas where data are not readily available, especially if referring to a discussion of future technologies, which do not yet exist. DEA, on the other hand, provides an ex post analysis of the past from which to learn. A discussion of this topic can be found in Belton and Stewart (1999). 9. Illustration For the benefit of the reader a simple example has been analyzed using the majority of techniques presented in this review. The data has been drawn from the nursing home example developed in Sexton et al. (1986), in which six DMUs are compared over four variables: staff hours per day (StHr), supplies per day (Supp), total Medicare plus Medicaid reimbursed patient days (MCPD) and total private patient days (PPPD). The raw data are presented in Table 1 and the results of the CCR, BCC and additive models are presented in Table 2.

12 260 N. Adler et al. / European Journal of Operational Research 140 (2002) Table 1 Raw data of nursing home example Inputs Outputs StHr Supp MCPD PPPD A B C D E F As can be seen in Table 2, whilst both the CCR and additive models define DMUs E and F as inefficient, the BCC model, which allows for variable returns-to-scale, only defines F as inefficient. Table 3 presents the rankings of the DMUs according to the cross-efficiency and super-efficiency models. As can be seen in Table 3, the results of the cross-efficiency models according to the aggressive and benevolent objective functions are different one from the other, although A is clearly the most efficient whilst C and F are in the worst relative positions. In the aggressive CE model, no DMU achieves an average efficiency score of one, whilst in the benevolent CE model, both units A and D do achieve an average efficient score of one. The super-efficient model ranks A as the most efficient and E and F as inefficient (since the latter matches the CCR model). Table 4 presents the results of the benchmarking and MID rankings, with the former ranking only the efficient units (defined according to the variable Table 2 DMU scores for standard three DEA models CCR BCC Additive constant returns-to-scale A 1.00 A 1.00 A 0.00 B 1.00 B 1.00 B 0.00 C 1.00 C 1.00 C 0.00 D 1.00 D 1.00 D 0.00 E 0.98 E 1.00 E F 0.87 F 0.90 F Table 3 Results of cross-efficiency and super-efficiency constant returns-to-scale ranks Cross-efficiency aggressive Cross-efficiency benevolent Super-efficiency A A 1 A B D 1 B D E C E B D C C E F F F Table 4 Benchmarking and MID ranks Benchmarking Rank A 1 E 0.99 B 2 F 0.93 E 3 C 4.5 D 4.5 MID (based on slacks from additive constant returns-to-scale model)

13 N. Adler et al. / European Journal of Operational Research 140 (2002) returns-to-scale additive model in this example) and the latter ranking the inefficient units alone. Once again, unit A achieves the top ranking in the benchmarking approach, whilst units C and D achieve the lowest ranking, amongst efficient DMUs. In the CE models, D is always ranked higher than C, and the opposite is true in the super-efficient model. Clearly, the logic behind the reason for ranking the DMUs will decide the ultimate ranking procedure chosen and consequently the results. The MID model ranks unit E higher than F, as occurred in the three basic DEA models, but gives them a higher relative efficient score. Table 5 presents the results of the statistical models developed in Section 6. All three statistical model results rank the DMUs differently, although notably F is always ranked last. Furthermore, A loses its first ranking in the DDEA model alone, possibly due to the common weights approach, since this can also be seen in the maximin efficiency ratio model with common weights presented in Table 6. This table presents the results of the MCDM models of Troutt (1995) and Li and Reeves (1999). According to the maximin common weights Table 5 Ranking according to statistical-based models DMU Rank CCA DDEA DR/DEA A B C D E F model, DMU D is ranked first, whereas the MOLP objective functions define both units A and D as relatively efficient, in a similar manner to that of the benevolent CE model. The advantage of MOLP is the evaluation of a usable set of weights per DMU within which discussion as to improvements can be developed, which is somewhat missing from the cross-efficiency approach. 10. Conclusions The field of data envelopment analysis has grown at an exponential rate (see Seiford, 1996) since the seminal papers of Farrell (1957) and Charnes et al. (1978). The original idea of evaluating after-school programs with multiple inputs and outputs has led to an enormous body of academic literature. Within this field is a sub-group of papers in which many researchers have sought to improve the differential capabilities of DEA and to fully rank both efficient, as well as inefficient, decision-making units. In this review, the DEA ranking concepts have been divided into six general areas. The first group of papers is based on a cross-efficiency matrix. By evaluating DMUs through both self- and peer pressure, one can attain a more balanced view of the decision-making units. The second group of papers is based on the super-efficiency approach, in which the efficient units can receive a score greater than one, through the unit s exclusion from the column being scored in the linear program. This proved popular and many papers sprouted from this idea, broadening its use from mere ranking to outlier detection, sensitivity analyses and scale classification. However, each unit is Table 6 Ranking according to MCDM models Maximin efficiency ratio MOLP minimax MOLP minisum D A 1 A 1 C D 1 D 1 E E E A B F B C B F F C 0.830

14 262 N. Adler et al. / European Journal of Operational Research 140 (2002) evaluated by its own weights as opposed to the cross-efficiency concept in which all units are compared using the same sets of weights. The third grouping is based on benchmarking, in which a DMU is highly ranked if it is chosen as a useful target for many other DMUs. This is of substantial use when looking to benchmark industries. The fourth group of papers developed a connection between multivariate statistical techniques and DEA. Canonical correlation analysis and discriminant analysis were each used to compute common weights, from which the set of DMUs can be ranked. In practice, non-parametric statistical tests showed a strong correlation between the final ranking and the original DEA dichotomous classification. The fifth section discussed the ranking of inefficient units. One approach, entitled a measure of inefficiency dominance, ranks the inefficient units according to their average proportional inefficiency in all inputs and outputs. In the last set of papers, which crosses multi-criteria decision-making models with DEA, some concepts used additional, preferential information in order to aid the ranking process. The additional information can be incorporated into or alongside the standard DEA results through the use of fuzzy logic, assurance regions or discrimination intensity functions. Other concepts combined the two approaches without the need for additional information such as AHP, the maximin efficiency ratio model and a multi-objective linear program. It should be noted that many papers have been written in an empirical context, utilizing the concepts discussed here. However, referring to all the published applications in this area would result in an overly long review hence they were considered beyond the scope of this review. Our aim was to review the methodologies rather than the subsequent applications, though it can be noted that certain techniques have been heavily used in spe- Table 7 Criteria for ranking the main methodologies Ranking only efficient units Ranking by new concept (not h k ) Cross-efficiency models Sexton et al. (1986) a One set of common weights Occasional problems with infeasibility Super-efficiency models Andersen and Petersen (1993) True for Hashimoto (1997) efficient Mehrabian et al. (1999) units only Sueyoshi (1999) Benchmarking Torgersen et al. (1996) Statistics-based models Friedman and Sinuany-Stern (1997) Sinuany-Stern et al. (1994) Sinuany-Stern and Friedman (1998) Friedman and Sinuany-Stern (1998) Ranking of inefficient units Bardhan et al. (1996) MCDA/DEA Troutt (1995) Li and Reeves (1999) Karsak (1998) Hougaard (1999) Sinuany-Stern et al. (2000) a The cross-efficiency models use n sets of common weights applied to all DMUs.

15 N. Adler et al. / European Journal of Operational Research 140 (2002) Table 8 DEA software packages and their ranking capabilities Super-efficiency Cross-efficiency Benchmark (simple count) Statistics DEAP a DEA-Solver-Pro b EMS c Frontier d IDEAS e On Front f Warwick g a b See c EMS is free to the academic community and available from Holger Scheel at d See e See f See g Contact e.thanassoulis@aston.ac.uk. cific areas. Cross-efficiency has been applied in many areas of manufacturing, including engineering design, flexible manufacturing systems, industrial robotics and business process re-engineering etc. It has also been used heavily in project and R&D portfolio selection. Super-efficiency has been applied in a wide range of papers from financial institutions and industry to public regulation, education and healthcare. Benchmarking has been used extensively in the field of utilities, industry and agricultural productivity. The statistical techniques have been applied to universities and industry and MCDA/DEA to agriculture and oil. Clearly, these methodologies have a wideranging applicability in many areas of both the public and private sectors. To aid in choosing any of these concepts we have summarized the six major ranking methods according to various criteria in Table 7. In addition, the commercial DEA software manufacturers were contacted and asked to specify the ranking techniques available in their programs. Seven software programs were known to the authors and the results are as shown in Table 8. Evidently, the super-efficient and benchmarking approaches are the most widely spread. Finally, many mathematical and statistical techniques have been presented here, all with the basic aim of increasing the discriminating power of data envelopment analysis and ranking the decision-making units. However, whilst each technique may be useful in a specific area, no one methodology can be prescribed here as the panacea of all ills. It remains to be seen whether the ultimate DEA model can be developed to solve all problems and which will consequently be easy to solve by practitioners in the field and academics alike. The trend, however, would appear to be in this direction, as evidenced by Cooper et al. (1999) in which many ideas have been incorporated into one model. The model includes the capability of dealing with imprecise data as well as assurance region and cone-ratio concepts. Another potentially fruitful concept may be to invoke several ranking procedures and then to compute an average or median rank based on all the models employed, as suggested in Friedman and Sinuany-Stern (1998). This concept, used frequently by economists, statisticians and forecasters, may help to achieve a complete ranking whilst avoiding some of the pitfalls of each individual approach. References Adler, N., Golany, B., Evaluation of deregulated airline networks using data envelopment analysis combined with principal component analysis with an application to Western Europe. European Journal of Operational Research 132 (2), Andersen, P., Petersen, N.C., A procedure for ranking efficient units in data envelopment analysis. Management Science 39 (10),

Further discussion on linear production functions and DEA

Further discussion on linear production functions and DEA European Journal of Operational Research 127 (2000) 611±618 www.elsevier.com/locate/dsw Theory and Methodology Further discussion on linear production functions and DEA Joe Zhu * Department of Management,

More information

Classifying inputs and outputs in data envelopment analysis

Classifying inputs and outputs in data envelopment analysis European Journal of Operational Research 180 (2007) 692 699 Decision Support Classifying inputs and outputs in data envelopment analysis Wade D. Cook a, *, Joe Zhu b a Department of Management Science,

More information

A Slacks-base Measure of Super-efficiency for Dea with Negative Data

A Slacks-base Measure of Super-efficiency for Dea with Negative Data Australian Journal of Basic and Applied Sciences, 4(12): 6197-6210, 2010 ISSN 1991-8178 A Slacks-base Measure of Super-efficiency for Dea with Negative Data 1 F. Hosseinzadeh Lotfi, 2 A.A. Noora, 3 G.R.

More information

Review of Methods for Increasing Discrimination in Data Envelopment Analysis

Review of Methods for Increasing Discrimination in Data Envelopment Analysis Annals of Operations Research 116, 225 242, 2002 2002 Kluwer Academic Publishers. Manufactured in The Netherlands. Review of Methods for Increasing Discrimination in Data Envelopment Analysis LIDIA ANGULO-MEZA

More information

A DEA- COMPROMISE PROGRAMMING MODEL FOR COMPREHENSIVE RANKING

A DEA- COMPROMISE PROGRAMMING MODEL FOR COMPREHENSIVE RANKING Journal of the Operations Research Society of Japan 2004, Vol. 47, No. 2, 73-81 2004 The Operations Research Society of Japan A DEA- COMPROMISE PROGRAMMING MODEL FOR COMPREHENSIVE RANKING Akihiro Hashimoto

More information

Data envelopment analysis

Data envelopment analysis 15 Data envelopment analysis The purpose of data envelopment analysis (DEA) is to compare the operating performance of a set of units such as companies, university departments, hospitals, bank branch offices,

More information

Equivalent Standard DEA Models to Provide Super-Efficiency Scores

Equivalent Standard DEA Models to Provide Super-Efficiency Scores Second Revision MS 5035 Equivalent Standard DEA Models to Provide Super-Efficiency Scores C.A.K. Lovell 1 and A.P.B. Rouse 2 1 Department of Economics 2 Department of Accounting and Finance Terry College

More information

Chapter 2 Output Input Ratio Efficiency Measures

Chapter 2 Output Input Ratio Efficiency Measures Chapter 2 Output Input Ratio Efficiency Measures Ever since the industrial revolution, people have been working to use the smallest effort to produce the largest output, so that resources, including human,

More information

Searching the Efficient Frontier in Data Envelopment Analysis INTERIM REPORT. IR-97-79/October. Pekka Korhonen

Searching the Efficient Frontier in Data Envelopment Analysis INTERIM REPORT. IR-97-79/October. Pekka Korhonen IIASA International Institute for Applied Systems Analysis A-2361 Laxenburg Austria Tel: +43 2236 807 Fax: +43 2236 71313 E-mail: info@iiasa.ac.at Web: www.iiasa.ac.at INTERIM REPORT IR-97-79/October Searching

More information

A DIMENSIONAL DECOMPOSITION APPROACH TO IDENTIFYING EFFICIENT UNITS IN LARGE-SCALE DEA MODELS

A DIMENSIONAL DECOMPOSITION APPROACH TO IDENTIFYING EFFICIENT UNITS IN LARGE-SCALE DEA MODELS Pekka J. Korhonen Pyry-Antti Siitari A DIMENSIONAL DECOMPOSITION APPROACH TO IDENTIFYING EFFICIENT UNITS IN LARGE-SCALE DEA MODELS HELSINKI SCHOOL OF ECONOMICS WORKING PAPERS W-421 Pekka J. Korhonen Pyry-Antti

More information

AN IMPROVED APPROACH FOR MEASUREMENT EFFICIENCY OF DEA AND ITS STABILITY USING LOCAL VARIATIONS

AN IMPROVED APPROACH FOR MEASUREMENT EFFICIENCY OF DEA AND ITS STABILITY USING LOCAL VARIATIONS Bulletin of Mathematics Vol. 05, No. 01 (2013), pp. 27 42. AN IMPROVED APPROACH FOR MEASUREMENT EFFICIENCY OF DEA AND ITS STABILITY USING LOCAL VARIATIONS Isnaini Halimah Rambe, M. Romi Syahputra, Herman

More information

ABSTRACT INTRODUCTION

ABSTRACT INTRODUCTION Implementation of A Log-Linear Poisson Regression Model to Estimate the Odds of Being Technically Efficient in DEA Setting: The Case of Hospitals in Oman By Parakramaweera Sunil Dharmapala Dept. of Operations

More information

PRIORITIZATION METHOD FOR FRONTIER DMUs: A DISTANCE-BASED APPROACH

PRIORITIZATION METHOD FOR FRONTIER DMUs: A DISTANCE-BASED APPROACH PRIORITIZATION METHOD FOR FRONTIER DMUs: A DISTANCE-BASED APPROACH ALIREZA AMIRTEIMOORI, GHOLAMREZA JAHANSHAHLOO, AND SOHRAB KORDROSTAMI Received 7 October 2003 and in revised form 27 May 2004 In nonparametric

More information

Modeling undesirable factors in efficiency evaluation

Modeling undesirable factors in efficiency evaluation European Journal of Operational Research142 (2002) 16 20 Continuous Optimization Modeling undesirable factors in efficiency evaluation Lawrence M. Seiford a, Joe Zhu b, * www.elsevier.com/locate/dsw a

More information

Ranking Decision Making Units with Negative and Positive Input and Output

Ranking Decision Making Units with Negative and Positive Input and Output Int. J. Research in Industrial Engineering, pp. 69-74 Volume 3, Number 3, 2014 International Journal of Research in Industrial Engineering www.nvlscience.com Ranking Decision Making Units with Negative

More information

The Comparison of Stochastic and Deterministic DEA Models

The Comparison of Stochastic and Deterministic DEA Models The International Scientific Conference INPROFORUM 2015, November 5-6, 2015, České Budějovice, 140-145, ISBN 978-80-7394-536-7. The Comparison of Stochastic and Deterministic DEA Models Michal Houda, Jana

More information

Indian Institute of Management Calcutta. Working Paper Series. WPS No. 787 September 2016

Indian Institute of Management Calcutta. Working Paper Series. WPS No. 787 September 2016 Indian Institute of Management Calcutta Working Paper Series WPS No. 787 September 2016 Improving DEA efficiency under constant sum of inputs/outputs and Common weights Sanjeet Singh Associate Professor

More information

Joint Use of Factor Analysis (FA) and Data Envelopment Analysis (DEA) for Ranking of Data Envelopment Analysis

Joint Use of Factor Analysis (FA) and Data Envelopment Analysis (DEA) for Ranking of Data Envelopment Analysis Joint Use of Factor Analysis () and Data Envelopment Analysis () for Ranking of Data Envelopment Analysis Reza Nadimi, Fariborz Jolai Abstract This article combines two techniques: data envelopment analysis

More information

Data Envelopment Analysis and its aplications

Data Envelopment Analysis and its aplications Data Envelopment Analysis and its aplications VŠB-Technical University of Ostrava Czech Republic 13. Letní škola aplikované informatiky Bedřichov Content Classic Special - for the example The aim to calculate

More information

USING LEXICOGRAPHIC PARAMETRIC PROGRAMMING FOR IDENTIFYING EFFICIENT UNITS IN DEA

USING LEXICOGRAPHIC PARAMETRIC PROGRAMMING FOR IDENTIFYING EFFICIENT UNITS IN DEA Pekka J. Korhonen Pyry-Antti Siitari USING LEXICOGRAPHIC PARAMETRIC PROGRAMMING FOR IDENTIFYING EFFICIENT UNITS IN DEA HELSINKI SCHOOL OF ECONOMICS WORKING PAPERS W-381 Pekka J. Korhonen Pyry-Antti Siitari

More information

Sensitivity and Stability Radius in Data Envelopment Analysis

Sensitivity and Stability Radius in Data Envelopment Analysis Available online at http://ijim.srbiau.ac.ir Int. J. Industrial Mathematics Vol. 1, No. 3 (2009) 227-234 Sensitivity and Stability Radius in Data Envelopment Analysis A. Gholam Abri a, N. Shoja a, M. Fallah

More information

Finding the strong defining hyperplanes of production possibility set with constant returns to scale using the linear independent vectors

Finding the strong defining hyperplanes of production possibility set with constant returns to scale using the linear independent vectors Rafati-Maleki et al., Cogent Mathematics & Statistics (28), 5: 447222 https://doi.org/.8/233835.28.447222 APPLIED & INTERDISCIPLINARY MATHEMATICS RESEARCH ARTICLE Finding the strong defining hyperplanes

More information

INEFFICIENCY EVALUATION WITH AN ADDITIVE DEA MODEL UNDER IMPRECISE DATA, AN APPLICATION ON IAUK DEPARTMENTS

INEFFICIENCY EVALUATION WITH AN ADDITIVE DEA MODEL UNDER IMPRECISE DATA, AN APPLICATION ON IAUK DEPARTMENTS Journal of the Operations Research Society of Japan 2007, Vol. 50, No. 3, 163-177 INEFFICIENCY EVALUATION WITH AN ADDITIVE DEA MODEL UNDER IMPRECISE DATA, AN APPLICATION ON IAUK DEPARTMENTS Reza Kazemi

More information

A ratio-based method for ranking production units in profit efficiency measurement

A ratio-based method for ranking production units in profit efficiency measurement Math Sci (2016) 10:211 217 DOI 10.1007/s40096-016-0195-8 ORIGINAL RESEARCH A ratio-based method for ranking production units in profit efficiency measurement Roza Azizi 1 Reza Kazemi Matin 1 Gholam R.

More information

A New Group Data Envelopment Analysis Method for Ranking Design Requirements in Quality Function Deployment

A New Group Data Envelopment Analysis Method for Ranking Design Requirements in Quality Function Deployment Available online at http://ijim.srbiau.ac.ir/ Int. J. Industrial Mathematics (ISSN 2008-5621) Vol. 9, No. 4, 2017 Article ID IJIM-00833, 10 pages Research Article A New Group Data Envelopment Analysis

More information

Subhash C Ray Department of Economics University of Connecticut Storrs CT

Subhash C Ray Department of Economics University of Connecticut Storrs CT CATEGORICAL AND AMBIGUOUS CHARACTERIZATION OF RETUNS TO SCALE PROPERTIES OF OBSERVED INPUT-OUTPUT BUNDLES IN DATA ENVELOPMENT ANALYSIS Subhash C Ray Department of Economics University of Connecticut Storrs

More information

Identifying Efficient Units in Large-Scale Dea Models

Identifying Efficient Units in Large-Scale Dea Models Pyry-Antti Siitari Identifying Efficient Units in Large-Scale Dea Models Using Efficient Frontier Approximation W-463 Pyry-Antti Siitari Identifying Efficient Units in Large-Scale Dea Models Using Efficient

More information

Author Copy. A modified super-efficiency DEA model for infeasibility. WD Cook 1,LLiang 2,YZha 2 and J Zhu 3

Author Copy. A modified super-efficiency DEA model for infeasibility. WD Cook 1,LLiang 2,YZha 2 and J Zhu 3 Journal of the Operational Research Society (2009) 60, 276 --281 2009 Operational Research Society Ltd. All rights reserved. 0160-5682/09 www.palgrave-journals.com/jors/ A modified super-efficiency DEA

More information

Optimal weights in DEA models with weight restrictions

Optimal weights in DEA models with weight restrictions Loughborough University Institutional Repository Optimal weights in DEA models with weight restrictions This item was submitted to Loughborough University's Institutional Repository by the/an author. Citation:

More information

Data Envelopment Analysis within Evaluation of the Efficiency of Firm Productivity

Data Envelopment Analysis within Evaluation of the Efficiency of Firm Productivity Data Envelopment Analysis within Evaluation of the Efficiency of Firm Productivity Michal Houda Department of Applied Mathematics and Informatics Faculty of Economics, University of South Bohemia in Česé

More information

Selective Measures under Constant and Variable Returns-To- Scale Assumptions

Selective Measures under Constant and Variable Returns-To- Scale Assumptions ISBN 978-93-86878-06-9 9th International Conference on Business, Management, Law and Education (BMLE-17) Kuala Lumpur (Malaysia) Dec. 14-15, 2017 Selective Measures under Constant and Variable Returns-To-

More information

Groups performance ranking based on inefficiency sharing

Groups performance ranking based on inefficiency sharing Available online at http://ijim.srbiau.ac.ir/ Int. J. Industrial Mathematics (ISSN 2008-5621) Vol. 5, No. 4, 2013 Article ID IJIM-00350, 9 pages Research Article Groups performance ranking based on inefficiency

More information

NEGATIVE DATA IN DEA: A SIMPLE PROPORTIONAL DISTANCE FUNCTION APPROACH. Kristiaan Kerstens*, Ignace Van de Woestyne**

NEGATIVE DATA IN DEA: A SIMPLE PROPORTIONAL DISTANCE FUNCTION APPROACH. Kristiaan Kerstens*, Ignace Van de Woestyne** Document de travail du LEM 2009-06 NEGATIVE DATA IN DEA: A SIMPLE PROPORTIONAL DISTANCE FUNCTION APPROACH Kristiaan Kerstens*, Ignace Van de Woestyne** *IÉSEG School of Management, CNRS-LEM (UMR 8179)

More information

Preference aggregation and DEA: An analysis of the methods proposed to discriminate efficient candidates

Preference aggregation and DEA: An analysis of the methods proposed to discriminate efficient candidates Preference aggregation and DEA: An analysis of the methods proposed to discriminate efficient candidates Bonifacio Llamazares, Teresa Peña Dep. de Economía Aplicada, PRESAD Research Group, Universidad

More information

Chance Constrained Data Envelopment Analysis The Productive Efficiency of Units with Stochastic Outputs

Chance Constrained Data Envelopment Analysis The Productive Efficiency of Units with Stochastic Outputs Chance Constrained Data Envelopment Analysis The Productive Efficiency of Units with Stochastic Outputs Michal Houda Department of Applied Mathematics and Informatics ROBUST 2016, September 11 16, 2016

More information

Determination of Economic Optimal Strategy for Increment of the Electricity Supply Industry in Iran by DEA

Determination of Economic Optimal Strategy for Increment of the Electricity Supply Industry in Iran by DEA International Mathematical Forum, 2, 2007, no. 64, 3181-3189 Determination of Economic Optimal Strategy for Increment of the Electricity Supply Industry in Iran by DEA KH. Azizi Department of Management,

More information

CENTER FOR CYBERNETIC STUDIES

CENTER FOR CYBERNETIC STUDIES TIC 1 (2 CCS Research Report No. 626 Data Envelopment Analysis 1% by (A. Charnes W.W. Cooper CENTER FOR CYBERNETIC STUDIES DTIC The University of Texas Austin,Texas 78712 ELECTE fl OCT04109W 90 - CCS Research

More information

CHAPTER 3 THE METHOD OF DEA, DDF AND MLPI

CHAPTER 3 THE METHOD OF DEA, DDF AND MLPI CHAPTER 3 THE METHOD OF DEA, DDF AD MLP 3.1 ntroduction As has been discussed in the literature review chapter, in any organization, technical efficiency can be determined in order to measure the performance.

More information

Measuring the efficiency of assembled printed circuit boards with undesirable outputs using Two-stage Data Envelopment Analysis

Measuring the efficiency of assembled printed circuit boards with undesirable outputs using Two-stage Data Envelopment Analysis Measuring the efficiency of assembled printed circuit boards with undesirable outputs using Two-stage ata Envelopment Analysis Vincent Charles CENTRUM Católica, Graduate School of Business, Pontificia

More information

Special Cases in Linear Programming. H. R. Alvarez A., Ph. D. 1

Special Cases in Linear Programming. H. R. Alvarez A., Ph. D. 1 Special Cases in Linear Programming H. R. Alvarez A., Ph. D. 1 Data Envelopment Analysis Objective To compare technical efficiency of different Decision Making Units (DMU) The comparison is made as a function

More information

European Journal of Operational Research

European Journal of Operational Research European Journal of Operational Research 207 (200) 22 29 Contents lists available at ScienceDirect European Journal of Operational Research journal homepage: www.elsevier.com/locate/ejor Interfaces with

More information

Sensitivity and Stability Analysis in DEA on Interval Data by Using MOLP Methods

Sensitivity and Stability Analysis in DEA on Interval Data by Using MOLP Methods Applied Mathematical Sciences, Vol. 3, 2009, no. 18, 891-908 Sensitivity and Stability Analysis in DEA on Interval Data by Using MOLP Methods Z. Ghelej beigi a1, F. Hosseinzadeh Lotfi b, A.A. Noora c M.R.

More information

EFFICIENCY ANALYSIS UNDER CONSIDERATION OF SATISFICING LEVELS

EFFICIENCY ANALYSIS UNDER CONSIDERATION OF SATISFICING LEVELS EFFICIENCY ANALYSIS UNDER CONSIDERATION OF SATISFICING LEVELS FOR OUTPUT QUANTITIES [004-0236] Malte L. Peters, University of Duisburg-Essen, Campus Essen Stephan Zelewski, University of Duisburg-Essen,

More information

A reference-searching based algorithm for large-scale data envelopment. analysis computation. Wen-Chih Chen *

A reference-searching based algorithm for large-scale data envelopment. analysis computation. Wen-Chih Chen * A reference-searching based algorithm for large-scale data envelopment analysis computation Wen-Chih Chen * Department of Industrial Engineering and Management National Chiao Tung University, Taiwan Abstract

More information

A Data Envelopment Analysis Based Approach for Target Setting and Resource Allocation: Application in Gas Companies

A Data Envelopment Analysis Based Approach for Target Setting and Resource Allocation: Application in Gas Companies A Data Envelopment Analysis Based Approach for Target Setting and Resource Allocation: Application in Gas Companies Azam Mottaghi, Ali Ebrahimnejad, Reza Ezzati and Esmail Khorram Keywords: power. Abstract

More information

Symmetric Error Structure in Stochastic DEA

Symmetric Error Structure in Stochastic DEA Available online at http://ijim.srbiau.ac.ir Int. J. Industrial Mathematics (ISSN 2008-5621) Vol. 4, No. 4, Year 2012 Article ID IJIM-00191, 9 pages Research Article Symmetric Error Structure in Stochastic

More information

Interior-Point Methods for Linear Optimization

Interior-Point Methods for Linear Optimization Interior-Point Methods for Linear Optimization Robert M. Freund and Jorge Vera March, 204 c 204 Robert M. Freund and Jorge Vera. All rights reserved. Linear Optimization with a Logarithmic Barrier Function

More information

PRIORITIZATION METHOD FOR FRONTIER DMUs: A DISTANCE-BASED APPROACH

PRIORITIZATION METHOD FOR FRONTIER DMUs: A DISTANCE-BASED APPROACH PRIORITIZATION METHOD FOR FRONTIER DMUs: A DISTANCE-BASED APPROACH ALIREZA AMIRTEIMOORI, GHOLAMREZA JAHANSHAHLOO, AND SOHRAB KORDROSTAMI Received 7 October 2003 and in revised form 27 May 2004 In nonparametric

More information

A METHOD FOR SOLVING 0-1 MULTIPLE OBJECTIVE LINEAR PROGRAMMING PROBLEM USING DEA

A METHOD FOR SOLVING 0-1 MULTIPLE OBJECTIVE LINEAR PROGRAMMING PROBLEM USING DEA Journal of the Operations Research Society of Japan 2003, Vol. 46, No. 2, 189-202 2003 The Operations Research Society of Japan A METHOD FOR SOLVING 0-1 MULTIPLE OBJECTIVE LINEAR PROGRAMMING PROBLEM USING

More information

LITERATURE REVIEW. Chapter 2

LITERATURE REVIEW. Chapter 2 Chapter 2 LITERATURE REVIEW This chapter reviews literature in the areas of productivity and performance measurement using mathematical programming and multi-level planning for resource allocation and

More information

MULTI-COMPONENT RETURNS TO SCALE: A DEA-BASED APPROACH

MULTI-COMPONENT RETURNS TO SCALE: A DEA-BASED APPROACH Int. J. Contemp. Math. Sci., Vol. 1, 2006, no. 12, 583-590 MULTI-COMPONENT RETURNS TO SCALE: A DEA-BASED APPROACH Alireza Amirteimoori 1 Department of Mathematics P.O. Box: 41335-3516 Islamic Azad University,

More information

PRINCIPAL COMPONENT ANALYSIS TO RANKING TECHNICAL EFFICIENCIES THROUGH STOCHASTIC FRONTIER ANALYSIS AND DEA

PRINCIPAL COMPONENT ANALYSIS TO RANKING TECHNICAL EFFICIENCIES THROUGH STOCHASTIC FRONTIER ANALYSIS AND DEA PRINCIPAL COMPONENT ANALYSIS TO RANKING TECHNICAL EFFICIENCIES THROUGH STOCHASTIC FRONTIER ANALYSIS AND DEA Sergio SCIPPACERCOLA Associate Professor, Department of Economics, Management, Institutions University

More information

Sensitivity and Stability Analysis in Uncertain Data Envelopment (DEA)

Sensitivity and Stability Analysis in Uncertain Data Envelopment (DEA) Sensitivity and Stability Analysis in Uncertain Data Envelopment (DEA) eilin Wen a,b, Zhongfeng Qin c, Rui Kang b a State Key Laboratory of Virtual Reality Technology and Systems b Department of System

More information

Prediction of A CRS Frontier Function and A Transformation Function for A CCR DEA Using EMBEDED PCA

Prediction of A CRS Frontier Function and A Transformation Function for A CCR DEA Using EMBEDED PCA 03 (03) -5 Available online at www.ispacs.com/dea Volume: 03, Year 03 Article ID: dea-0006, 5 Pages doi:0.5899/03/dea-0006 Research Article Data Envelopment Analysis and Decision Science Prediction of

More information

Investigación Operativa. New Centralized Resource Allocation DEA Models under Constant Returns to Scale 1

Investigación Operativa. New Centralized Resource Allocation DEA Models under Constant Returns to Scale 1 Boletín de Estadística e Investigación Operativa Vol. 28, No. 2, Junio 2012, pp. 110-130 Investigación Operativa New Centralized Resource Allocation DEA Models under Constant Returns to Scale 1 Juan Aparicio,

More information

Selecting most efficient information system projects in presence of user subjective opinions: a DEA approach

Selecting most efficient information system projects in presence of user subjective opinions: a DEA approach CEJOR https://doi.org/10.1007/s10100-018-0549-4 ORIGINAL PAPER Selecting most efficient information system projects in presence of user subjective opinions: a DEA approach Mehdi Toloo 1 Soroosh Nalchigar

More information

A MODIFIED DEA MODEL TO ESTIMATE THE IMPORTANCE OF OBJECTIVES WITH AN APPLICATION TO AGRICULTURAL ECONOMICS

A MODIFIED DEA MODEL TO ESTIMATE THE IMPORTANCE OF OBJECTIVES WITH AN APPLICATION TO AGRICULTURAL ECONOMICS A MODIFIED DEA MODEL TO ESTIMATE THE IMPORTANCE OF OBJECTIVES WITH AN APPLICATION TO AGRICULTURAL ECONOMICS Francisco J. André Inés Herrero Laura Riesgo Pablo de Olavide University Ctra. de Utrera, km.

More information

Research Article A Data Envelopment Analysis Approach to Supply Chain Efficiency

Research Article A Data Envelopment Analysis Approach to Supply Chain Efficiency Advances in Decision Sciences Volume 2011, Article ID 608324, 8 pages doi:10.1155/2011/608324 Research Article A Data Envelopment Analysis Approach to Supply Chain Efficiency Alireza Amirteimoori and Leila

More information

Partial Input to Output Impacts in DEA: Production Considerations and Resource Sharing among Business Subunits

Partial Input to Output Impacts in DEA: Production Considerations and Resource Sharing among Business Subunits Partial Input to Output Impacts in DEA: Production Considerations and Resource Sharing among Business Subunits Raha Imanirad, 1 Wade D. Cook, 1 Joe Zhu 2 1 Operations Management and Information Systems

More information

15-780: LinearProgramming

15-780: LinearProgramming 15-780: LinearProgramming J. Zico Kolter February 1-3, 2016 1 Outline Introduction Some linear algebra review Linear programming Simplex algorithm Duality and dual simplex 2 Outline Introduction Some linear

More information

CHAPTER 4 VARIABILITY ANALYSES. Chapter 3 introduced the mode, median, and mean as tools for summarizing the

CHAPTER 4 VARIABILITY ANALYSES. Chapter 3 introduced the mode, median, and mean as tools for summarizing the CHAPTER 4 VARIABILITY ANALYSES Chapter 3 introduced the mode, median, and mean as tools for summarizing the information provided in an distribution of data. Measures of central tendency are often useful

More information

Measures of Central Tendency and their dispersion and applications. Acknowledgement: Dr Muslima Ejaz

Measures of Central Tendency and their dispersion and applications. Acknowledgement: Dr Muslima Ejaz Measures of Central Tendency and their dispersion and applications Acknowledgement: Dr Muslima Ejaz LEARNING OBJECTIVES: Compute and distinguish between the uses of measures of central tendency: mean,

More information

Linear Programming in Matrix Form

Linear Programming in Matrix Form Linear Programming in Matrix Form Appendix B We first introduce matrix concepts in linear programming by developing a variation of the simplex method called the revised simplex method. This algorithm,

More information

Transformation ofvariablesin Data EnvelopmentAnalysis

Transformation ofvariablesin Data EnvelopmentAnalysis DepartmentofInformation andserviceeconomy On theuseofanon-singularlinear Transformation ofvariablesin Data EnvelopmentAnalysis AbolfazlKeshvari Pekka Korhonen BUSINESS+ ECONOMY WORKINGPAPERS Aalto University

More information

A Fuzzy Data Envelopment Analysis Approach based on Parametric Programming

A Fuzzy Data Envelopment Analysis Approach based on Parametric Programming INT J COMPUT COMMUN, ISSN 1841-9836 8(4:594-607, August, 013. A Fuzzy Data Envelopment Analysis Approach based on Parametric Programming S.H. Razavi, H. Amoozad, E.K. Zavadskas, S.S. Hashemi Seyed Hossein

More information

CHAPTER 4 MEASURING CAPACITY UTILIZATION: THE NONPARAMETRIC APPROACH

CHAPTER 4 MEASURING CAPACITY UTILIZATION: THE NONPARAMETRIC APPROACH 49 CHAPTER 4 MEASURING CAPACITY UTILIZATION: THE NONPARAMETRIC APPROACH 4.1 Introduction: Since the 1980's there has been a rapid growth in econometric studies of capacity utilization based on the cost

More information

Using AHP for Priority Determination in IDEA

Using AHP for Priority Determination in IDEA Applied Mathematical Sciences, Vol. 5, 2011, no. 61, 3001-3010 Using AHP for Priority Determination in IDEA Gholam Reza Jahanshahloo Department of Mathematics, Science and Research Branch Islamic Azad

More information

USING STRATIFICATION DATA ENVELOPMENT ANALYSIS FOR THE MULTI- OBJECTIVE FACILITY LOCATION-ALLOCATION PROBLEMS

USING STRATIFICATION DATA ENVELOPMENT ANALYSIS FOR THE MULTI- OBJECTIVE FACILITY LOCATION-ALLOCATION PROBLEMS USING STRATIFICATION DATA ENVELOPMENT ANALYSIS FOR THE MULTI- OBJECTIVE FACILITY LOCATION-ALLOCATION PROBLEMS Jae-Dong Hong, Industrial Engineering, South Carolina State University, Orangeburg, SC 29117,

More information

Marginal values and returns to scale for nonparametric production frontiers

Marginal values and returns to scale for nonparametric production frontiers Loughborough University Institutional Repository Marginal values and returns to scale for nonparametric production frontiers This item was submitted to Loughborough University's Institutional Repository

More information

Department of Social Systems and Management. Discussion Paper Series

Department of Social Systems and Management. Discussion Paper Series Department of Social Systems and Management Discussion Paper Series No. 1128 Measuring the Change in R&D Efficiency of the Japanese Pharmaceutical Industry by Akihiro Hashimoto and Shoko Haneda July 2005

More information

Data Envelopment Analysis with metaheuristics

Data Envelopment Analysis with metaheuristics Data Envelopment Analysis with metaheuristics Juan Aparicio 1 Domingo Giménez 2 José J. López-Espín 1 Jesús T. Pastor 1 1 Miguel Hernández University, 2 University of Murcia ICCS, Cairns, June 10, 2014

More information

Combining DEA and factor analysis to improve evaluation of academic departments given uncertainty about the output constructs.

Combining DEA and factor analysis to improve evaluation of academic departments given uncertainty about the output constructs. Combining DEA and factor analysis to improve evaluation of academic departments given uncertainty about the output constructs. SClaudina Vargas and Dennis Bricker 1 Department of Industrial Engineering,

More information

FACTOR ANALYSIS AND MULTIDIMENSIONAL SCALING

FACTOR ANALYSIS AND MULTIDIMENSIONAL SCALING FACTOR ANALYSIS AND MULTIDIMENSIONAL SCALING Vishwanath Mantha Department for Electrical and Computer Engineering Mississippi State University, Mississippi State, MS 39762 mantha@isip.msstate.edu ABSTRACT

More information

Revenue Malmquist Index with Variable Relative Importance as a Function of Time in Different PERIOD and FDH Models of DEA

Revenue Malmquist Index with Variable Relative Importance as a Function of Time in Different PERIOD and FDH Models of DEA Revenue Malmquist Index with Variable Relative Importance as a Function of Time in Different PERIOD and FDH Models of DEA Mohammad Ehsanifar, Golnaz Mohammadi Department of Industrial Management and Engineering

More information

Data Envelopment Analysis (DEA) with an applica6on to the assessment of Academics research performance

Data Envelopment Analysis (DEA) with an applica6on to the assessment of Academics research performance Data Envelopment Analysis (DEA) with an applica6on to the assessment of Academics research performance Outline DEA principles Assessing the research ac2vity in an ICT School via DEA Selec2on of Inputs

More information

Testing the Validity of the Travel and Tourism Competitiveness Index in the World Economic Forum with Classical and Fuzzy Data Envelopment Analyses

Testing the Validity of the Travel and Tourism Competitiveness Index in the World Economic Forum with Classical and Fuzzy Data Envelopment Analyses PTiR 2013;4;121-128 121 Hatice Ozkoc, Hakan Bakan, Ercan Baldemi r Muğla Sitki Koçman Üniversitesi, Turkey Testing the Validity of the Travel and Tourism Competitiveness Index in the World Economic Forum

More information

Using Genetic Algorithms for Maximizing Technical Efficiency in Data Envelopment Analysis

Using Genetic Algorithms for Maximizing Technical Efficiency in Data Envelopment Analysis Using Genetic Algorithms for Maximizing Technical Efficiency in Data Envelopment Analysis Juan Aparicio 1 Domingo Giménez 2 Martín González 1 José J. López-Espín 1 Jesús T. Pastor 1 1 Miguel Hernández

More information

Centre for Efficiency and Productivity Analysis

Centre for Efficiency and Productivity Analysis Centre for Efficiency and Productivity Analysis Working Paper Series No. WP04/2018 Profit Efficiency, DEA, FDH and Big Data Valentin Zelenyuk Date: June 2018 School of Economics University of Queensland

More information

A buyer - seller game model for selection and negotiation of purchasing bids on interval data

A buyer - seller game model for selection and negotiation of purchasing bids on interval data International Mathematical Forum, 1, 2006, no. 33, 1645-1657 A buyer - seller game model for selection and negotiation of purchasing bids on interval data G. R. Jahanshahloo, F. Hosseinzadeh Lotfi 1, S.

More information

UC Riverside UC Riverside Previously Published Works

UC Riverside UC Riverside Previously Published Works UC Riverside UC Riverside Previously Published Works Title On the Relationship among Values of the same Summary Measure of Error when used across Multiple Characteristics at the same point in time: An

More information

Lecture Note 18: Duality

Lecture Note 18: Duality MATH 5330: Computational Methods of Linear Algebra 1 The Dual Problems Lecture Note 18: Duality Xianyi Zeng Department of Mathematical Sciences, UTEP The concept duality, just like accuracy and stability,

More information

Finding Closest Targets in Non-Oriented DEA Models: The Case of Convex and Non-Convex Technologies

Finding Closest Targets in Non-Oriented DEA Models: The Case of Convex and Non-Convex Technologies Finding Closest Targets in Non-Oriented DEA Models: The Case of Convex and Non-Convex Technologies MARIA CONCEIÇÃO A. SILVA PORTELA csilva@porto.ucp.pt; portemca@aston.ac.uk Universidade Católica Portuguesa

More information

CHAPTER 8 INTRODUCTION TO STATISTICAL ANALYSIS

CHAPTER 8 INTRODUCTION TO STATISTICAL ANALYSIS CHAPTER 8 INTRODUCTION TO STATISTICAL ANALYSIS LEARNING OBJECTIVES: After studying this chapter, a student should understand: notation used in statistics; how to represent variables in a mathematical form

More information

REMOVING INCONSISTENCY IN PAIRWISE COMPARISON MATRIX IN THE AHP

REMOVING INCONSISTENCY IN PAIRWISE COMPARISON MATRIX IN THE AHP MULTIPLE CRITERIA DECISION MAKING Vol. 11 2016 Sławomir Jarek * REMOVING INCONSISTENCY IN PAIRWISE COMPARISON MATRIX IN THE AHP DOI: 10.22367/mcdm.2016.11.05 Abstract The Analytic Hierarchy Process (AHP)

More information

A Simple Characterization

A Simple Characterization 95-B-1 A Simple Characterization of Returns to Scale in DEA l(aoru Tone Graduate School of Policy Science Saitama University Urawa, Saita1na 338, Japan tel: 81-48-858-6096 fax: 81-48-852-0499 e-mail: tone@poli-sci.saita1na-u.ac.jp

More information

Mixed input and output orientations of Data Envelopment Analysis with Linear Fractional Programming and Least Distance Measures

Mixed input and output orientations of Data Envelopment Analysis with Linear Fractional Programming and Least Distance Measures STATISTICS, OPTIMIZATION AND INFORMATION COMPUTING Stat., Optim. Inf. Comput., Vol. 4, December 2016, pp 326 341. Published online in International Academic Press (www.iapress.org) Mixed input and output

More information

Chapter 2 Network DEA Pitfalls: Divisional Efficiency and Frontier Projection

Chapter 2 Network DEA Pitfalls: Divisional Efficiency and Frontier Projection Chapter 2 Network DEA Pitfalls: Divisional Efficiency and Frontier Projection Yao Chen, Wade D. Cook, Chiang Kao, and Joe Zhu Abstract Recently network DEA models have been developed to examine the efficiency

More information

CS281 Section 4: Factor Analysis and PCA

CS281 Section 4: Factor Analysis and PCA CS81 Section 4: Factor Analysis and PCA Scott Linderman At this point we have seen a variety of machine learning models, with a particular emphasis on models for supervised learning. In particular, we

More information

Math for Machine Learning Open Doors to Data Science and Artificial Intelligence. Richard Han

Math for Machine Learning Open Doors to Data Science and Artificial Intelligence. Richard Han Math for Machine Learning Open Doors to Data Science and Artificial Intelligence Richard Han Copyright 05 Richard Han All rights reserved. CONTENTS PREFACE... - INTRODUCTION... LINEAR REGRESSION... 4 LINEAR

More information

A New Fuzzy Positive and Negative Ideal Solution for Fuzzy TOPSIS

A New Fuzzy Positive and Negative Ideal Solution for Fuzzy TOPSIS A New Fuzzy Positive and Negative Ideal Solution for Fuzzy TOPSIS MEHDI AMIRI-AREF, NIKBAKHSH JAVADIAN, MOHAMMAD KAZEMI Department of Industrial Engineering Mazandaran University of Science & Technology

More information

Fuzzy efficiency: Multiplier and enveloping CCR models

Fuzzy efficiency: Multiplier and enveloping CCR models Available online at http://ijim.srbiau.ac.ir/ Int. J. Industrial Mathematics (ISSN 28-5621) Vol. 8, No. 1, 216 Article ID IJIM-484, 8 pages Research Article Fuzzy efficiency: Multiplier and enveloping

More information

Chapter 2 An Overview of Multiple Criteria Decision Aid

Chapter 2 An Overview of Multiple Criteria Decision Aid Chapter 2 An Overview of Multiple Criteria Decision Aid Abstract This chapter provides an overview of the multicriteria decision aid paradigm. The discussion covers the main features and concepts in the

More information

Indicator: Proportion of the rural population who live within 2 km of an all-season road

Indicator: Proportion of the rural population who live within 2 km of an all-season road Goal: 9 Build resilient infrastructure, promote inclusive and sustainable industrialization and foster innovation Target: 9.1 Develop quality, reliable, sustainable and resilient infrastructure, including

More information

Data envelopment analysis approach for discriminating efficient candidates in voting systems by considering the priority of voters

Data envelopment analysis approach for discriminating efficient candidates in voting systems by considering the priority of voters Data envelopment analysis approach for discriminating efficient candidates in voting systems by considering the priority of voters Ali Ebrahimnejad Keywords: candidate. Abstract There are different ways

More information

Income elasticity of human development in ASEAN countries

Income elasticity of human development in ASEAN countries The Empirical Econometrics and Quantitative Economics Letters ISSN 2286 7147 EEQEL all rights reserved Volume 2, Number 4 (December 2013), pp. 13-20. Income elasticity of human development in ASEAN countries

More information

SENSITIVITY ANALYSIS IN LINEAR PROGRAMING: SOME CASES AND LECTURE NOTES

SENSITIVITY ANALYSIS IN LINEAR PROGRAMING: SOME CASES AND LECTURE NOTES SENSITIVITY ANALYSIS IN LINEAR PROGRAMING: SOME CASES AND LECTURE NOTES Samih Antoine Azar, Haigazian University CASE DESCRIPTION This paper presents case studies and lecture notes on a specific constituent

More information

Logistic Regression: Regression with a Binary Dependent Variable

Logistic Regression: Regression with a Binary Dependent Variable Logistic Regression: Regression with a Binary Dependent Variable LEARNING OBJECTIVES Upon completing this chapter, you should be able to do the following: State the circumstances under which logistic regression

More information

Exponential smoothing in the telecommunications data

Exponential smoothing in the telecommunications data Available online at www.sciencedirect.com International Journal of Forecasting 24 (2008) 170 174 www.elsevier.com/locate/ijforecast Exponential smoothing in the telecommunications data Everette S. Gardner

More information

STAT Section 5.8: Block Designs

STAT Section 5.8: Block Designs STAT 518 --- Section 5.8: Block Designs Recall that in paired-data studies, we match up pairs of subjects so that the two subjects in a pair are alike in some sense. Then we randomly assign, say, treatment

More information

DECISIONS UNDER UNCERTAINTY

DECISIONS UNDER UNCERTAINTY August 18, 2003 Aanund Hylland: # DECISIONS UNDER UNCERTAINTY Standard theory and alternatives 1. Introduction Individual decision making under uncertainty can be characterized as follows: The decision

More information