We  deal with the relationship termination problem in the context of individual-level customer relationship management (CRM) and use a Markov decision process to determine the most appropriate occasion for termination of the relationship with a seemingly unprofitable customer. As a particular case, the beta-geometric/beta-binomial model is considered as the basis to define customer behavior and it is explained how to compute customer lifetime value when one needs to take account of the firm’s choice as to whether to continue or terminate relationship with unprofitable customers. By numerical examples provided by simulation, it is shown how a stochastic dynamic programming approach can be adopted in order to obtain a more precise estimation of the customer lifetime value as a key criterion for resource allocation in CRM.     

  Since late 1960's, the emergency location problems, fire stations and medical emergency services have attracted the attention of researchers. Mathematical models, both deterministic and probabilistic, have been proposed and applied to find suitable locations for such facilities in many urban and rural areas. Here, we review some models proposed for finding the location of such facilities, with an eye on successfully implemented real life applications. We then propose an extension of the QM-CLAM model of Marianov and Serra (1998) to M/G/k systems, and suggest a GRASP type heuristic procedure for solving the problem. To improve the computed solution, local search heuristics are used. Sensitivity analysis and some computational results are also presented.

  We relax some assumptions of the traditional scheduling problem and suggest an adapted meta-heuristic algorithm to optimize efficient utilization of resources and quick response to demands simultaneously. We intend to bridge the existing gap between theory and real industrial scheduling assumptions (e.g., hot metal rolling industry, chemical and pharmaceutical industries). We adapt and evaluate a well-known algorithm based on electromagnetic science. The motivation behind our proposed meta-heuristic approach has arisen from the attraction-repulsion mechanism of electromagnetic theories in physics. In this basic idea, we desire to bring our search closer to a region with a superior objective function while going away from the region with the inferior objective function in order to move the solution gradually towards optimality. The algorithm is carefully evaluated for its performance against two existing algorithms using multi-objective performance measures and statistical tools. The results show that our proposed solution method outperforms the others.

  We consider criticality in project networks having imprecise activity duration times. It is well known that finding all possibly critical paths of an imprecise project network is an NP-hard problem. Here, based on a method for finding critical paths of crisp networks by using only the forward recursion of critical path method, for the first time an algorithm is proposed which can find all possibly critical paths of interval-valued project networks. The proposed algorithm considers interactivity among paths which has not been yet considered in the fuzzy project scheduling literature. The extension of the proposed algorithm to the fuzzy network calculates criticality degrees of activities and paths of projects without any need to enumerate all project paths. Although algorithms for calculating criticality degrees in fuzzy networks have been previously proposed, despite the fact that they mostly consider a specific type of fuzzy numbers as activity duration times, the exiting algorithms do not discriminate possibly critical paths before calculating the criticality degrees. The computational experience on a series of well-known project samples confirms the algorithm to be remarkably more efficient than similar algorithms for fuzzy networks.

  Among various statistical and data mining discriminant analysis proposed so far for group classification, linear programming discriminant analysis has recently attracted the researchers’ interest. This study evaluates multi-group discriminant linear programming (MDLP) for classification problems against well-known methods such as neural networks and support vector machine. MDLP is less complicated as compared to other methods and does not suffer from having local optima. This study also proposes a fuzzy Delphi method to select and gather the required data, when databases suffer from deficient data. In addition, to absorb the uncertainty infused to collecting data, interval MDLP (IMDLP) is developed. The results show that the performance of MDLP and specially IMDLP is better than conventional classification methods with respect to correct classification, at least for small and medium-size datasets.