Modern Research in Decision Making

Abstract Tobin index is one of the most important indices in the world of investment used as a criterion for evaluating performance of the firms to decide for the right investments. However, there are some ambiguities in the accuracy of the results based on this index that have prompted researchers to pursue estimation of this index based on the other financial indices. But Tobin index is a dynamic index and as it is based on the market price, may be changed its value at once, therefore it is not logical to be predicted using methods as multiple regression that attempt to predict precise value of depent variable. this research has reviewed methods based on the exact prediction like regression to judge about Tobin index by the other financial indices and it has recommended using discriminant analysis methods such as logistic regression and artificial nervous network. Discriminant analysis is a method to categorize a set of the observations into one of two or several determined groups, so that observations within each group can have the most similarity to each other. Therefore, this research has analyzed Tobin index discrimination using financial information of 184 accepted firms by Tehran Stock Exchange in the financial year leading to 29 of Esfand in 1393 by logistic regression and artificial nervous network and it has reported results of two techniques and has compared output of the two techniques to each other.

Abstract One of the popular problems in operations research and project management is resource-constrained project scheduling problem. In the present study, this problem is modeled considering important goals consisted of minimization of the project completion time, as well as minimization of the maximum cost of the project in one day. In this regard, all the possible prerequisite relations between the activities of a project are considered. The proposed model has been implemented for three real projects in different sizes and by using metaheuristic algorithms including genetic algorithm, particle swarm optimization and differential evolution. The results showed that differential evolution and particle swarm optimization algorithms have efficient performances compared to the genetic algorithms for large- and medium-scale projects respectively. The use of metaheuristic algorithms for solving small-scale projects is not recommended.

Abstract Upon equipping the banks to Electronic payment and receiving systems as well as the use of credit cards, most of the customers do their bank transactions by using credit cards and through the use of credit channels such as ATM machines, POS sale terminals, phone banking, internet banking, etc. The customers are now better able to find their required services and products and they may even change their own bank because of the type of services required. Therefore, managing customer relations is inevitable for banks. One of the helpful instruments in managing customer relations is data mining. Four data mining methods including decision tree, simple Biz, Vicinity neighbor K and combinatory model were used in this study to identify the most profitable services used by the customers. Each of these methods has been investigated on real data and the efficiency of each method has been examines. The results of model evaluations showed that vicinity neighbor K’s accuracy in finding the profitable services was equal to 93.26%, that of Biz was 74.83% and that of decision tree was 97.18%. in addition, the accuracy of combinatory model was 94.80%. Further, the combinatory model was successful in accurately identifying 96.01% of the normal services and it was also successful to accurately identify 94.44% of the services. Therefore, we may conclude that it has a far better performance as compared with Biz model and Vicinity Neighbor K. the evaluations results showed that combinatory model is more accurate to use as compared with other existing models.

Abstract The increasing trend of supply chain disruptions in the recent decades has increased the importance of considering resilience strategies in supply chain network design/redesign problems. The present study provides a scenario-based mixed integer two-stage stochastic programming model for dealing with operational and disruption risks in a retail supply chain network redesign problem. The mode covers both pre-disruption and post-disruption decisions jointly. Five resilience strategies, including proactive, reactive, and chain design quality aspects (i.e. facility fortification, safety stock inventory holding, using excess capacity at some critical nodes of the network, using the direct to store delivery as the shipping strategy and multiple covering of retail stores), are considered. The model is applied for a case study in the retail industry. Several sensitivity analyses are carried out from which useful managerial insights are drawn to be used by top managers. The results show the positive effect of applying resilience strategies on the total cost reduction.

Abstract In service organizations, improving service quality is critical for increasing productivity, profitability and customer satisfaction, and without identifying these criteria, improving service quality is not possible. Therefore, the aim of this paper is to introduce a Delphi method based on the concept of qualitative reasoning for identification and localization the effective criteria in the quality of services. Firstly, the concepts of the qualitative absolute order of magnitude method were described, and based it, a regular structure for the Delphi method was presented. In a qualitative reasoning environment, a consensus function by using the concept of entropy was introduced. Then, the mechanism for achieving convergence in the Delphi method was provided. According to the conceptual model offered by this research, identification and localization criteria for service quality of the public transportation in ShahreKurd city were investigated as a case study. The results showed that among the 35 initial criteria that were collected based on the combination of criteria in library studies and Expert Panel views, 30 criteria were selected as effective criteria to improve the service quality of Shahrekurd public transportation.

Abstract Location theory is an interstice field of optimization and operations research. In the classic location problem, the goal is finding the location of one or more facilities such that some criteria such as transportation cost, the sum of distances passed by clients, total service time and cost of servicing are minimized. In this paper, we consider the goal location problem. In the goal location problem, the ideal is locating the facility in the distances ri, from the i-th client. However, in the most instances, the solution of this problem doesn’t exist. Therefore, we consider the minimizing of distances between clients and ideal point. The minimizing sum of square errors and minimizing absolute errors under Lp norm are considered as the objective function. We use the Weiszfeld like, Gauss-Newton and imperialist competitive algorithms for solving the problem. Then we compare the results which obtained by these methods for some test problems.

Abstract Seru systems are among the most advanced repetitive manufacturing systems which produce in low or medium volumes.   These systems consist of several cells and get the best flow time through the new methods of assigning operators. In the literature, more attention is paid to the aspects of the cells formation and allocation and sequencing of the various products into the cells, and the assignment of operators, which is considered to be the most important element in these systems, has not been studied in line with other decisions. Therefore, in this paper, these decisions have been studied independently and analyzed through intra and inter serus job rotation scheduling. The presented ILP model assigns a given set of operators to the cells so that the total number of stays in a cell in successive rotation periods be minimized. GAMS software has been used to solve the model and also, the teaching and learning based optimization algorithm has been designed to improve the efficiency for problems of medium and large sizes. Several test problems have been generated and solved in a variety of sizes to examine the validity of the model and the performance of the algorithm. Results show the proper efficiency of the algorithm and quality of its solutions.

Abstract In previous studies, the design of the vendor managed inventory (VMI) models was based on minimizing the total cost of the supply chain. In this paper, a new approach for designing VMI models with the aim of optimizing the inventory turnover of the producer warehouse is developed. Based on the proposed model, our objectives are to maximize inventory turnover along with the constraint of lack of shortage of goods in the production lines and also compliance with the minimum and maximum constraints of inventory in the warehouse of the producer which can be simpler and more practical than minimizing the total cost of the supply chain. A hybrid algorithm based on a genetic algorithm (GA) using particle swarm optimization (PSO) is proposed in order to gain both proper global and local search abilities in the solution space for solving the new model. As a case study, implementation of the proposed model in the supply chain of Saipa Company improved the inventory turnover, decreased inventory level and decreased the level of replenishment by suppliers.

Abstract In knowledge base economy, performance evaluation of STPs as one of the most important entities in this economy is vital. In this research, we used literature review and industries expert interviews to determine what criteria are necessary to assess STPs. different topologies was used to assess the validity of the model and best model was chosen through highest reliability criteria. By comparison we made sure that selected criteria are reliably aligned with sience and technology parks objectives. On mathematical side we used robust multi-level DEA to assess STPs performances. We developed a Multi-level DEA with Non-Discretionary using banker's extension. Applying the developed model and the original model we assessed the efficiency of Iranian STPs.Then we compared the results for these two model on efficiency of DMUs, and their respecting weights using regression analysis. Based on the results two moderator variables (Park maturity and percentage of gross production in state) variable was identified and their impact on model and efficiency of DMUs was evaluated as high.