Stepwise selection of variables in data envelopment analysis : Procedures and managerial perspectives

Two-sided matching for hesitant fuzzy numbers in smart intelligent technique transfer

In an earlier work (Kamada and Kojima 2015), we introduced "matching with constraints," a two-sided matching problem in which the market is subject to feasibility constraints That paper proposed and analyzed a possible solution to this problem and discussed various real-market applications of our results The present paper reorganizes some of the findings from our more recent effort, based on Kamada and Kojima (2016a, b)

Recent Developments in Matching with Constraints

In ride-sharing, a sustainable transportation mode, drivers share idle seats with riders. This paper addresses two key subproblems to find a ride-sharing solution: matching drivers and riders to achieve the system-wide optimality and allocating the resulting travel-cost saving among them by considering their self interests or the stability of the solution. Solutions to either subproblem significantly affect the efficiency of overall dynamic ride-sharing systems measured in total travel-cost saving and individual benefits of people involved. We evaluate the worst-case performance of the best stable solutions associated with four ridematching-defined pricing methods existing in the literature and widely used in practice. We then analyze a Vickrey-Clarke-Groves (VCG)-based pricing mechanism that lets the system achieve optimal system efficiency, prove its stability, and identify its drawbacks such as budget unbalance. We formulate the matching subproblem to maximize the total travel-cost saving and propose a simple and consistent pricing method to guarantee that the resulting ride-sharing solution not only is stable and system-wide optimal but also makes the system deficit-free and thus financially sustainable. Furthermore, such a ride-sharing solution can be obtained in polynomial time. In the numerical experiments, we evaluate our method against existing ones through a real-life instance and randomly generated instances in a dynamic environment.

Matching and pricing in ride-sharing: Optimality, stability, and financial sustainability

The main task of a scheduler applied to astronomical observatories is the time optimization and the maximization of the scientific return. Scheduling of observations is an example of the classical task allocation problem known as the job-shop problem (JSP) or the flexible-JSP (fJSP). In most cases various mathematical algorithms are usually considered to solve the planning system. We present an analysis of the task allocation problem and the solutions currently in use at different astronomical facilities. We also describe the schedulers for three different projects (TJO, CARMENES and CTA) where the conclusions of this analysis are applied in their development.

Research on schedulers for astronomical observatories

Assignment of trainees to software project requirements

With the evolving data regulations and residency laws, it is becoming challenging for enterprises to serve all the geographically distributed users. To abide with the regulations and to leverage the benefits, such as good quality of service at a low cost, and avoid vendor lock-in, enterprises are seeking multiple Cloud Service Providers (CSPs) for hosting their web applications. The selection of CSPs in a multi-cloud environment is a challenging problem as it depends on multiple criteria such as security, service cost, manageability and so on.In this paper, we address the CSP selection problem for multiple applications in a multi-cloud environment comprising of multiple criteria. We provide a holistic solution methodology that ranks the CSP combinations for each web application using a multiple-criteria decision-making (MCDM) technique. The solution methodology takes virtual machine and cloud center selection into consideration for ranking the CSP combinations. However, selecting the top ranked CSP combinations for all applications may not be a viable option from the manageability and budget perspectives. Hence, we propose an optimization model to select CSPs for multiple applications of an enterprise considering budget and rank. We demonstrate our methodology through numerical experiments and provide various insights.

Selection of Cloud Service Providers for Hosting Web Applications in a Multi-cloud Environment

In this paper, we study a centralized matching scheme that has to assign a set of agents to a set of jobs by meeting distributional constraints. The scheme has to maximize social welfare and fairness offered by the matching to the parties. Furthermore, the allocation needs to minimize the number of blocking pairs. The problem is NP-hard and hence computationally challenging. Nonetheless, the users expect good solutions that can be generated quickly. We propose linear programming-based improvement heuristics to solve this multi-criteria stable matching problem. Our approach finds an equitable and global welfare stable matching solution in significantly lesser time. We then demonstrate the applicability of our proposed model and the solution methodology in a workforce allocation problem faced by software projects.

Multi-objective stable matching and distributional constraints

To evaluate performance of organizational units using Human Capital Value (HCV), the input data spread across an enterprise may be received from a user. The input data includes employee data, project related data, and organizational unit data for performance evaluation. The input data is analyzed for generating HCV variables. The HCV variables are stored in a repository ( 108 ). Further, the HCV variables may be parsed to determine an optimal set of variables. Based on the parsing, an efficiency of each organizational unit is computed. The computing is based on a Multi Criteria Decision Analysis (MCDA) technique. Based on the computing, the organizational units are ranked in a decreasing order of efficiency.

Evaluating performance of organizational units using human capital values

In this paper, we study a centralized, stable matching scheme, which allocates trainees to software project requirements, to minimize retraining and relocation costs when the preference lists of the project requirements may contain ties of arbitrary lengths. This particular trainees' assignment problem is important because the allocation decisions not only influence the costs but also impact software project deliverables and intern attrition rates. It is also an NP-hard problem because of the inclusion of the ties, and the costs in the stable allocation model. We, therefore, have designed a GRASP-based scatter search method, to solve the large size instances of our assignment problem efficiently. The GRASP method uses randomized algorithms to generate initial trial solutions. A repair heuristic based on regret minimization idea is designed to convert an unstable solution to a stable solution during an improvement phase. Computational experiments suggest that the proposed algorithm significantly reduces run time compared to the CPLEX, and produces solutions that are at an average 4.5% away from the best CPLEX solutions for the large size problem instances. Moreover, our scatter search method consistently provides better quality solutions than the two state of the art methods from the prior literature.

Mangesh Gharote

Papers

Assignment of trainees to software project requirements

Selection of Cloud Service Providers for Hosting Web Applications in a Multi-cloud Environment

Multi-objective stable matching and distributional constraints

Evaluating performance of organizational units using human capital values

Scatter search for trainees to software project requirements stable allocation