Asia-Pacific Journal of Operational Research

The data envelopment analysis (DEA) method is a mathematical programming approach to evaluate the relative performance of portfolios. Considering that the risk input indicators of existing DEA performance evaluation indices cannot reflect the pervasive fat tails and asymmetry in return distributions of mutual funds, we originally introduce new risk measures CVaR and VaR into inputs of relevant DEA indices to measure relative performance of portfolios more objectively. To fairly evaluate the performance variation of the same fund during different time periods, we creatively treat them as different decision making units (DMUs). Different from available DEA applications which mainly investigate the American mutual fund performance from the whole market or industry aspect, we analyze in detail the effect of different input/output indicator combinations on the performance of individual funds. Our empirical results show that VaR and CVaR, especially their combinations with traditional risk measures, are very helpful for comprehensively describing return distribution properties such as skewness and leptokurtosis, and can thus better evaluate the overall performance of mutual funds.

The lecture timetabling problem is known to be a highly constrained combinatorial optimization problem. There have been many attempts to address this problem using integer programming, graph coloring and several heuristic search methods. However, since each university has its own timetable setting requirements, it is difficult to develop a general solution method. Thus, the work is generally done manually. This paper attempts to solve the lecture timetabling problem of the University of Asmara using a customized memetic algorithm that we have called ALTUMA. It is a hybrid of genetic algorithms with hill-climbing operators. The performance of ALTUMA was evaluated using data obtained from the University. Empirical results show that ALTUMA is capable of producing good results in a reasonable amount of time. Besides, the results demonstrate that incorporating local search operators with a probabilistic scheme and delta method of fitness evaluation into the memetic algorithm significantly improves the search capabilities of the algorithm.

This paper considers a 2-station-5-class re-entrant line with infinite supply of work. We obtain the necessary and sufficient condition for its corresponding fluid model to be stable. As an application of the result, the positive Harris recurrence of the 2-station-5-class re-entrant line is established via the fluid model.

Using a dynamic programming formulation, an analysis is presented of both the first and second innings of a one-day cricket match assuming variation in type of ball bowled and subsequent selection of a strategy by the batsman. We assume that the team batting first uses the strategy to maximize the expected score, and the team batting second uses the strategy to maximize the probability of outscoring the first team and thus of maximizing the probability of a win. The dynamic programming formulation allows a calculation, at any stage of the innings, of the optimal scoring strategy depending on the type of ball bowled, along with an estimate of the maximum of the expected number of runs scored in the remainder of the first innings, and the maximum probability of a win in the second innings. Modifications are then introduced to examine the effect of tailender batsmen and a "fifth bowler". Finally a simulation is done to estimate the variance in total score by following the optimal strategy used in the first innings.

This paper investigates the impact of learning curve effect on setup cost for the continuous review inventory model involving controllable lead time with the mixture of backorder and partial lost sales. A learning curve is a well known tool which describes the relation between the performance of a task and the number of repetitions of that task. The objective of this study is to minimize the expected total annual cost by simultaneously optimizing order quantity, safety factor and lead time under different setup learning rates. There are two models considered in the paper, one with normal demand distribution and another with general demand distribution having both mean and variance known and finite. Numerical examples are presented to illustrate the procedures of the proposed solution algorithms, along with the savings on the total costs of the models with the inclusion of the learning effect on setup.

Multi-server, finite buffer, performance models of queueing systems are very useful tools for manufacturing, telecommunication, transportation and facility modelling applications. Exact computation of performance measures for general service multi-server queueing systems remains an intractable problem. Approximations of these performance measures are important to quickly and accurately reveal the performance of a system. This is desirable for both performance evaluation as well as optimization of these systems. Two-moment approximation formulas are presented for performance modelling of multi-server systems involving servers of 2,3,…,10 servers. Extensive computational results are provided to evaluate the approximation results against simulation, known tabular results, and other approximation formulas. Applications of the model to optimizing manufacturing and service systems using a marginal allocation algorithm are briefly illustrated. Extensions of the two-moment methodology to larger multi-server systems c = {25, 50, 100} round out the paper.

In this paper we propose a method for finding the reference set of a decision making unit (DMU), without chasing down all alternative optimal solutions of the envelopment form, which is a strong degenerate problem. The reference set is useful as a benchmark for an inefficient DMU, for identifying the status of returns to scale, ranking of DMUs and so on. Lastly, numerical examples are shown to illustrate our proposed approach.

Given an undirected, connected, weighted graph, the leaf-constrained minimum spanning tree (LCMST) problem seeks a spanning tree of the graph with smallest weight among all spanning trees of the graph, which contains at least l leaves. In this paper we have proposed two new metaheuristic approaches for the LCMST problem. One is an ant-colony optimization (ACO) algorithm, whereas the other is a tabu search based algorithm. Similar to a previously proposed genetic algorithm, these metaheuristic approaches also use the subset coding that represents a leaf-constrained spanning tree by the set of its interior vertices. Our new approaches perform well in comparison with two best heuristics reported in the literature for the problem — the subset-coded genetic algorithm and a greedy heuristic.