Publications without Fulltext
Permanent URI for this collectionhttps://hdl.handle.net/20.500.14288/3
Browse
12 results
Search Results
Publication Metadata only Optimizing digital twin synchronization in a finite horizon(IEEE, 2022) Matta, Andrea; Department of Business Administration; Tan, Barış; Department of Business Administration; College of Administrative Sciences and EconomicsGiven the tendency to increase the complexity of digital twins to capture a manufacturing system in the most detailed way, synchronizing and using a complex digital twin with the real-time data may require significant resources. We define the optimal synchronization problem to operate the digital twins in the most effective way by balancing the trade-off between improving the accuracy of the simulation prediction and using more resources. We formulate and solve the optimal synchronization problem for a special case. We analyze the characteristics of the state-dependent and state-independent optimal policies that indicate when to synchronize the simulation at each decision epoch. Our numerical experiments show that the number of synchronizations decreases with the synchronization cost and with the system variability. Furthermore, a lower average number of synchronizations can be achieved by using a state-dependent policy.Publication Metadata only Optimal timing of insulin initiation using a mathematical model(IEEE, 2019) Minsin, F. Erkam; Dursun, Mehtap; Department of Business Administration; Güneş, Evrim Didem; Faculty Member; Department of Business Administration; College of Administrative Sciences and Economics; 51391Diabetes is a chronic disease that is rapidly spreading worldwide. In the treatment of patients with type 2 diabetes, the most common type of the disease, doctors have many intervention methods. Therefore, taking the right action at the most appropriate time among these many types of interventions will make a great contribution to the patient's health as well as the costs of treatment. In this study, we developed a finite-horizon Markov decision process (MDP) model in which the state of the process will cover the patient's metabolic health status and history of drug used to solve this problem. The model is operated by using existing data in the literature and the results are discussed.Publication Metadata only Dynamic churn prediction framework with more effective use of rare event data: the case of private banking(Pergamon-Elsevier Science Ltd, 2014) Department of Business Administration; N/A; Ali, Özden Gür; Arıtürk, Umut; Faculty Member; PhD Student; Department of Business Administration; College of Administrative Sciences and Economics; Graduate School of Business; 57780; N/ACustomer churn prediction literature has been limited to modeling churn in the next (feasible) time period. On the other hand, lead time specific churn predictions can help businesses to allocate retention efforts across time, as well as customers, and identify early triggers and indicators of customer churn. We propose a dynamic churn prediction framework for generating training data from customer records, and leverage it for predicting customer churn within multiple horizons using standard classifiers. Further, we empirically evaluate the proposed approach in a case study about private banking customers in a European bank. The proposed framework includes customer observations from different time periods, and thus addresses the absolute rarity issue that is relevant for the most valuable customer segment of many companies. It also increases the sampling density in the training data and allows the models to generalize across behaviors in different time periods while incorporating the impact of the environmental drivers. As a result, this framework significantly increases the prediction accuracy across prediction horizons compared to the standard approach of one observation per customer; even when the standard approach is modified with oversampling to balance the data, or lags of customer behavior features are added as additional predictors. The proposed approach to dynamic churn prediction involves a set of independently trained horizon-specific binary classifiers that use the proposed dataset generation framework. In the absence of predictive dynamic churn models, we had to benchmark survival analysis which is used predominantly as a descriptive tool. The proposed method outperforms survival analysis in terms of predictive accuracy for all lead times, with a much lower variability. Further, unlike Cox regression, it provides horizon specific ranking of customers in terms of churn probability which allows allocation of retention efforts across customers and time periods. (C) 2014 Elsevier Ltd. All rights reserved.Publication Metadata only Modeling and analysis of a cooperative service network(Pergamon-Elsevier Science Ltd, 2021) N/A; N/A; Department of Business Administration; Hosseini, Behnaz; Tan, Barış; PhD Student; Faculty Member; Department of Business Administration; Graduate School of Sciences and Engineering; College of Administrative Sciences and Economics; N/A; 28600With the advances in technology and changes in customers' attitude towards different service delivery formats, it is important for the service providers to deliver online services in addition to the traditional face-to-face services. In the cooperative service network presented in this study, service providers cooperate to serve online service requests received by the network in addition to their own customers. Designing and managing the cooperative network effectively increase the utilization of the involved servers, provide an adequate service for the external customers, and increase the profit for both the network and service providers. From the operational perspective, the number and utilization of the members to be included in the network and the price that will be paid to each member for a directed request are the main design questions. In order to answer these questions, we present a stochastic model that captures the dynamics of customer arrivals, assignment, and admission control. To establish this model, we first derive the solution of the dynamic admission control problem for the servers who decide how to admit their own customers and the external online customers using a Markov decision process. We then analyze the operation of the whole network with the servers who use the optimal admission control policy and obtain the system performance measures depending on the members' operational parameters. These results are used to determine the optimal number of servers in the network and the service price to be paid to the participating servers in order to maximize the obtained profit. We show that a cooperative service network is an effective way of utilizing the idle capacity of the servers while providing an adequate service level for the external online customers and increasing the profit for both the network and service providers.Publication Metadata only Evaluating average and heterogeneous treatment effects in light of domain knowledge: impact of behaviors on disease prevalence(Ieee, 2019) N/A; Department of Business Administration; Ghanem, Angi Nazih; Ali, Özden Gür; N/A; Faculty Member; Department of Business Administration; N/A; College of Administrative Sciences and Economics; N/A; 57780Understanding causal treatment effect and its heterogeneity can improve targeting of efforts for prevention and treatment of diseases. A number of methods are emerging to estimate heterogeneous treatment effect from observational data, such as Causal Forest. In this paper, we evaluate the heterogeneous treatment effect estimates in terms of whether they recover the expected direction of the effect based on domain knowledge. We use the individual level health surveys conducted by the Turkish Statistical Institute (TUIK) over the span of eight years with 90K+ observations. We estimate the effect of six behaviors on the probability of two diseases (IHD and Diabetes). We compare two approaches: a) treatment and disease specific Causal Forest models that directly estimate the heterogeneous treatment effect, and b) disease specific Random Forest models of disease probability that are used as simulators to evaluate counterfactual scenarios. We find that, with some exceptions, the signs of Causal Forest heterogeneous treatment effects are aligned with domain knowledge. Causal Forest performed better than the more naive approach of using RF models as simulators which disregards selection bias in treatment assignment.Publication Metadata only Stockout risk estimation and expediting for repairable spare parts(Pergamon-Elsevier Science Ltd, 2022) Hekimoğlu, Mustafa; Şahin, Mustafa; Department of Business Administration; Kök, Abdullah Gürhan; Faculty Member; Department of Business Administration; College of Administrative Sciences and Economics; 108423Stockouts of repairable spares usually lead to significant downtime costs. Managers of Maintenance Repair Organizations (MROs) seek advance indicators of future stockouts which might allow them to take proactive actions that are beneficial for achieving target service levels with reasonable costs. Among such (proactive) actions, the most common, and the cheapest one is expediting existing repair processes. In this study, we develop an advance stockout risk estimation system for repairable spare parts. To the best of our knowledge, this is the first study to estimate the future stockout risk of a repairable part. The method considers different statistics, e.g. the number of ongoing repair processes, demand rate, repair time, etc. to estimate stockout risk of a repairable part for a given planning horizon. In our field tests with empirical data, the suggested method overperforms two heuristic approaches and achieves accuracy rates of 63% for 15 day-planning horizon and 83% for 45 days. We also suggest a repairable inventory control system including repair expediting, inspection and con-demnation processes. To optimize the control parameters we suggest a simple algorithm considering two constraints: Target service level and maximum fraction of expedited demand. The algorithm is proved to be efficient for finding the optimum policy parameter in our tests with empirical data. Tests with empirical data suggest savings up to 8%. Both systems are implemented at an MRO as building blocks of a inventory control tower. The impact of the implementation is assessed with empirical simulations and verified from the financial indicators of the company.Publication Metadata only A stochastic model of vessel casualties resulting from oil tanker traffic through narrow waterways(Soc Computer Simulation, 1998) Otay, Emre; Department of Business Administration; Tan, Barış; Faculty Member; Department of Business Administration; College of Administrative Sciences and Economics; 28600In this paper, we present our preliminary results of a stochastic model to investigate vessel casualties resulting from tanker traffic through a waterway. A state-space model of the waterway is developed by dividing it into a number of grids. The model incorporates the effects of physical forcing mechanisms, i.e., currents and waves, winds, visibility, geometry of the waterway and the routes of individual vessels in the drift probabilities of the vessels. Then these probabilities are used as state-transition probabilities of a Markov chain. The transient analysis of the resulting time-varying Markov chain yields risk charts that show the casualty probabilities across the geometry of the waterway at a given time. Furthermore the steady-state analysis allows us to analyze the relationship between the vessel traffic intensity and a global measure of casualty risk.Publication Metadata only A near-optimal order-based inventory allocation rule in an assemble-to-order system and its applications to resource allocation problems(Springer, 2005) Xu, Susan Hong; Department of Business Administration; Akçay, Yalçın; Faculty Member; Department of Business Administration; College of Administrative Sciences and Economics; 51400Assemble-to-order (ATO) manufacturing strategy has taken over the more traditional make-to-stock (MTS) strategy in many high-tech firms. ATO strategy has enabled these firms to deliver customized demand timely and to benefit from risk pooling due to component commonality. However, multi-component, multi-product ATO systems pose challenging inventory management problems. In this chapter, we study the component allocation problem given a specific replenishment policy and realized customer demands. We model the problem as a general multidimensional knapsack problem (MDKP) and propose the primal effective capacity heuristic (PECH) as an effective and simple approximate solution procedure for this NP-hard problem. Although the heuristic is primarily designed for the component allocation problem in an ATO system, we suggest that it is a general solution method for a wide range of resource allocation problems. We demonstrate the effectiveness of the heuristic through an extensive computational study which covers problems from the literature as well as randomly generated instances of the general and 0-1 MDKP. In our study, we compare the performance of the heuristic with other approximate solution procedures from the ATO system and integer programming literature.Publication Metadata only The effects of brand equity on price strategies: an agent based model(Univ De La Laguna, 2009) Delre, Sebastiano A.; Department of Business Administration; Esseghaier, Skander; Faculty Member; Department of Business Administration; College of Administrative Sciences and Economics; N/AConsumers are highly sensible to different price structures and price promotions. Many studies have showed how custÖmers differently respond when prices are split into separate parts, e.g. a regular price and a shipping and handling surcharge. This phenomenon has recently received much more attention because online sales have continuously and substantially increased in the last years and because online sales imply a price partitioning: product price and shipping price. This gives opportunities to online retailers. They can decide whether to apply promotional tactics on both regular prices and on shipping and handling prices. The price partitioning decision becomes more complicated than usual. Retailers have to choose between a free shipping offer strategy and a price partitioning strategy. In the former case they have to decide a single price that includes the shipping cost and in the latter cases they have to choose upon two prices, a price for the item and a price for the shipping. This paper investigates how firms decide which of these two strategies to adopt and how their brand equities affect their decisions. An agent based model is built in order to replicate Gumus et al. (2009) results and to depart from it bringing new insights about market partitioning (how many firms adopt which strategy) and the effects of brand equities.Publication Metadata only A multi-start granular skewed variable neighborhood tabu search for the roaming salesman problem(Elsevier, 2021) Shahmanzari, Masoud; Department of Business Administration; Aksen, Deniz; Faculty Member; Department of Business Administration; College of Administrative Sciences and Economics; 40308This paper presents a novel hybrid metaheuristic algorithm for the Roaming Salesman Problem (RSP), called Multi-Start Granular Skewed Variable Neighborhood Tabu Search (MS-GSVNTS). The objective in RSP is to design daily tours for a traveling campaigner who collects rewards from activities in cities during a fixed planning horizon. RSP exhibits a number of exclusive features: It is selective which implies that not every node needs a visit. The rewards of cities are time-dependent. Daily tours can be either an open or a closed tour which implies the absence of a fixed depot. Instead, there is a campaign base that is to be attended frequently. Multiple visits are allowed for certain cities. The proposed method MS-GSVNTS is tested on 45 real-life instances from Turkey which are built with actual travel distances and times and on 10 large scale instances. Computational results suggest that MS-GSVNTS is superior to the existing solution methods developed for RSP. It produces 50 best known solutions including 18 ties and 32 new ones. The performance of MS-GSVNTS can be attributed to its multi-start feature, rich neighborhood structures, skewed moves, and granular neighborhoods.