Hekimoğlu, Mustafa

Hekimoğlu, Mustafa

Name Variants

Hekimoğlu, Mustafa
M.,Hekimoğlu
M. Hekimoğlu
Mustafa, Hekimoğlu
Hekimoglu, Mustafa
M.,Hekimoglu
M. Hekimoglu
Mustafa, Hekimoglu
Hekimoglu,M.
Hekimoglu, M.
Hekimoğlu, M.

Job Title

Doç. Dr.

Email Address

Mustafa.hekımoglu@khas.edu.tr

Full item page

Scholarly Output

30

Articles

21

Citation Count

0

Supervised Theses

3 Scholarly Output Search Results

Now showing 1 - 10 of 29

Citation Count: 7
Markov-Modulated Analysis of a Spare Parts System With Random Lead Times and Disruption Risks
(Elsevier Science Bv, 2018) Hekimoğlu, Mustafa; van der Laan, Ervin; Dekker, Rommert
Spare parts supply chains are highly dependent on the dynamics of their installed bases. A decreasing number of capital products in use increases the nonstationary supply-side risk especially towards the end-of-life of capital products. This supply-side risk appears to present itself through varying lead times coupled with supply disruptions. To model the nonstationary supply-side risk we consider an exogenous Markov chain that modulates random lead times and disruption probabilities. Assuming that order crossovers do not occur we prove the optimality of a state-dependent base stock policy. Later we conduct an impact study to understand the value of considering stochastic lead times and supply disruption risk in spare parts inventory control. Our results indicate that the coupled effect of random lead times and disruptions can be larger than the summation of individual effects even for moderate lead time variances. Also the effect of nonstationarity on total cost can be as large as the summation of all risk factors combined. In addition to this managerial insight we present a procedure for supply risk mitigation based on an empirical model and our mathematical model. Experiments on a real business case indicate that the procedure is capable of reducing costs while making the inventory system more prepared for disruptions. (C) 2018 Elsevier B.V. All rights reserved.
Demand Classification for Spare Parts Supply Chains in the Presence of Three Dimensional Printers
(Kadir Has Üniversitesi, 2022) İşler, Zülal; Hekimoğlu, Mustafa
Three-dimensional printers (3DPs) are currently the source of the supply chain and are used to ensure spare parts supply in case of shortages. However, the reliability of the part produced in 3DP is lower than the original part supplied by the original equipment manufacturer (OEM). Failure of parts creates demand and the failure probability of original and printed part is different than each other. Thus, knowing the total demand distribution have great importance in optimizing the order quantity given to the OEM in the presence of 3DPs. In this study, the demand distribution of system failures has been determined by using the distribution classification methods put forward by Ord (1967) and Adan et al. (1995). In line with the results, according to study of Ord(1967), demand distribution is found as Hypergeometric and Binomial distribution. Discrete distribution family of Adan et al. (1995) gives Binomial distribution for the system demand. All results are tested with chi-square test and likelihood ratio test.
Citation Count: 21
Evaluation of Water Supply Alternatives for Istanbul Using Forecasting and Multi-Criteria Decision Making Methods
(Elsevier Ltd, 2020) Savun Hekimoğlu, Başak; Erbay, Barbaros; Hekimoğlu, Mustafa; Burak, Selmin
Water scarcity is one of the most serious problems of the future due to increasing urbanization and water demand. Urban water planners need to balance increasing water demand with water resources that are under increasing pressure due to climate change and water pollution. Decision makers are forced to select the most appropriate water management alternative with respect to multiple, conflicting criteria based on short and long term projections of water demand in the future. In this paper, we consider water management in Istanbul, a megacity with a population of 15 million. Purpose: The purpose of this paper is to develop a method combining demand forecasting with multi-criteria decision making (MCDM) methods to evaluate five different water supply alternatives with respect to seven criteria using opinions of experts and stakeholders from different sectors. Methodology: To combine forecasting with MCDM, we design a data collection method in which we share our demand forecasts with our experts. For demand forecasting, we compare Holt-Winters, Seasonal Autoregressive Integrated Moving Average (S-ARIMA), and feedforward Artificial Neural Network (ANN) models and select S-ARIMA as the best forecasting model for monthly water consumption data. Generated demand projections are shared with experts from different sectors and collected data is evaluated with Fuzzy Theory using two distinct MCDM models: Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) and Preference Ranking Organization Method for Enrichment Evaluations (PROMETHEE). Also our analyses are complemented with two sensitivity analyses. Findings: Our results indicate that greywater reuse is the best alternative to satisfy the growing water demand of the city whereas all experts find desalination and inter-basin water transfer as the least attractive solutions. In addition, we adopt the PROMETHEE GDSS procedure to obtain a GAIA plane indicating consensus among experts. Furthermore, we find that our results are moderately sensitive to the number of experts and they are insensitive to changes in experts’ evaluations. Novelty: To the best of our knowledge, our study is the first one incorporating water demand and supply management concepts into the evaluation of alternatives. From a methodological perspective, water demand projections have never been used in an MCDM study in the literature. Also, this paper contributes to the literature with a mathematical construction of consensus and Monte Carlo simulations for the sufficiency of experts consulted in a study.
Citation Count: 0
Yedek Parçaların Talebe Yönelik Eklemeli Üretiminde Lazer Cilalamanın Optimum Karar Verme Politikası Üzerinde Etkisi
(2020) Hekimoğlu, Mustafa; Ulutan, Durul
Eklemeli imalatın yakınlarda bulunan bir 3D yazıcı kullanılarak sermaye ürünlerinin yedek parça ihtiyaçlarını karşılamak için kullanılması giderek yaygınlaşmaktadır. Böyle bir teknoloji, talebe-binaen parça üretimini mümkün kılarak arızaların rassallığı nedeniyle tutulan yedek parça envanterinin önemli bir kısmını ortadan kaldırma imkânı sunmaktadır. 3D yazıcı kullanımının en büyük sorunlarından biri olan basılı ve orijinal parçalar arasındaki kalite farkı, yüzey pürüzlülüğünü hafifleten ve ek maliyet terimi karşılığında parçaların güvenilirliğini artıran lazer parlatma kullanılarak azaltılabilir. Farklı parametreler kullanılarak, parçaların güvenilirliği, sermaye ürünlerinin ihtiyaçlarına ve sistemlerin durumuna göre değiştirilebilir. Bu çalışmada, basılı parçaların yüzey pürüzlülüğü ve güvenilirliğinin orijinal yedek parçaların envanter seviyeleri ile birlikte optimize edilmesi sorunu ele alınmıştır. Çalışmada, sınırlı bir planlama ufku üzerinde rastgele arızalara maruz kalan sabit sayıda özdeş makinadan oluşan bir üretim tesisi dikkate alınmıştır. Matematiksel analiz ve ayrıntılı sayısal deneyler kullanılarak, sistemin uygun maliyetli yönetimi için kritik olabilecek optimum kontrol politikası ve maliyet parametreleri arasındaki ilişki gösterilmiştir.
Citation Count: 2
Modeling Repair Demand in Existence of a Nonstationary Installed Base
(Elsevier, 2023) Hekimoglu, Mustafa; Karli, Deniz
Life cycles of products consist of 3 phases, namely growth, maturity, and decline phases. Modeling repair demand is particularly difficult in the growth and decline stages due to nonstationarity. In this study, we suggest respective stochastic models that capture the dynamics of repair demand in these two phases. We apply our theory to two different operations management problems. First, using the moments of spare parts demand, we suggest an algorithm that selects a parametric distribution from the hypergeometric family (Ord, 1967) for each period in time. We utilize the algorithm in a single echelon inventory control problem. Second, we focus on investment decisions of Original Equipment Manufacturers (OEMs) to extend economic lifetimes of products with technology upgrades. Our results indicate that the second moment is sufficient for growing customer bases, whereas using the third moment doubles the approximation quality of theoretical distributions for a declining customer base. From a cost minimization perspective, using higher moments of demand leads to savings up to 13.6% compared to the single-moment approach. Also, we characterize the optimal investment policy for lifetime extension decisions from risk-neutral and risk-averse perspectives. We find that there exists a critical level of investment cost and installed base size for profitability of lifetime extension for OEMs. From a managerial point of view, we find that a risk-neutral decision maker finds the lifetime extension problem profitable. In contrast, even a slight risk aversion can make the lifetime extension decision economically undesirable.
Citation Count: 1
Markdown Optimization in Apparel Retail Sector
(Springer Science and Business Media B.V., 2020) Yıldız, S.C.; Hekimoğlu, M.
Price discounts, known as markdowns, are important for fast fashion retailers to utilize inventory in a distribution channel using demand management. Estimating future demand for a given discount level requires the evaluation of historical sales data. In this evaluation recent observations might be more important than the older ones as majority of price discounts take place at the end of a selling season and that time period provides more accurate estimations. In this study, we consider a weighted least squares method for the parameter estimation of an empirical demand model used in a markdown optimization system. We suggest a heuristic procedure for the implementation of weighted least squares in a markdown optimization utilizing a generic weight function from the literature. We tested the suggested system using empirical data from a Turkish apparel retailer. Our results indicate that the weighted least squares method is more proper than the ordinary least squares for the fast fashion sales data as it captures price sensitivity of demand at the end of a selling season more accurately. © 2020, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG.
Citation Count: 0
A Comparative Application of Machine Learning Approaches To Win-Back Lost Customers
(Institute of Electrical and Electronics Engineers Inc., 2023) Yildirim, S.; Yucekaya, A.D.; Hekimoglu, M.; Ozcan, B.
Today's consumer is more knowledgeable and conscious than in the past. For this reason, it is quite possible for consumers to leave their service/product providers and start receiving service from another service/product provider. Without a recovery strategy, companies often do not target their lost disloyal customer portfolio correctly and encounter the problem of lost customers. Lost customers can cause loss both in economic terms and in terms of business potential. At the same time, lost customers can also be considered as profits given to rival companies. What if the companies could foresee lost customers who would not want to receive service from them again? Could companies win back their customers? At this point, the article proposes using machine learning methods to recover lost customers for service providers. The customers that are likely to be lost in the future are estimated using the article's past stories of an automotive company's lost customers. The data used is completely real. LGBM, XGBoost, and Random Forest methods were used to estimate lost customers. Finally, the authors select the machine learning with the highest predictive success for customer recovery and discuss why this method might have worked well. © 2023 IEEE.
Citation Count: 2
Optimum utilization of on-demand manufacturing and laser polishing in existence of supply disruption risk
(Elsevier, 2022) Ulutan, Durul; Isler, Zulal; Kaya, Burak Erkan; Hekimoglu, Mustafa
3D printing has moved from being a rapid prototyping tool to an additive manufacturing method within the last decade. Additive manufacturing can satisfy the need in dire situations where spare parts distribution is an issue but access to a 3D printer is much more likely and rapid than access to original parts. Managing inventories of spare parts can be tackled with more ease thanks to the reduced part types with additive manufacturing. While quality (in terms of reliability) of additively manufactured spare parts in terms of mechanical properties seem to be lower than original parts (particularly due to the inherent staircase appearance and the corresponding stress concentration zones that can lead to premature fatigue failure), use of post-processing subtractive techniques to correct such surface irregularities are found to improve reliability. While each process adds another layer of complexity to the cost minimization problem, demand uncertainty and risk of supply disruption represent the modern global problems faced recently. The problem tackled in this study is the joint optimization of the supply reliability considering the effect of laser polishing parameters and the demand uncertainty. In this problem, a condition of random breakdowns of identical products is considered. Also, the original supplier of machine components is subject to exogenous disruptions, such as strikes, raw material scarcity, or the COVID-19 pandemic. As a result, the optimum control policy with the right cost parameters was shown via numerical experiments originated from mathematical analyses. This optimality can be critical in managing the system in the best possible way, particularly during times of unforeseen circumstances such as pandemics. (C) 2022 Society of Manufacturing Engineers (SME). Published by Elsevier Ltd. All rights reserved. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the Scientific Committee of the NAMRI/SME.
Citation Count: 6
Optimization of Wastewater Treatment Systems for Growing Industrial Parks
(Elsevier B.V., 2023) Savun-Hekimoğlu, B.; İşler, Z.; Hekimoğlu, M.; Burak, S.; Karlı, D.; Yücekaya, A.; Akpınar, E.
Wastewater treatment is one of the crucial functions of industrial parks as wastewater from industrial facilities usually contains toxic compounds that can cause damage to the environment. To control their environmental loads, industrial parks make investment decisions for wastewater treatment plants. For this, they need to consider technical and economic factors as well as future growth projections as substantial construction and operational costs of wastewater treatment plants have to be shared by all companies in an industrial park. In this paper, we consider the long-term capacity planning problem for wastewater treatment facilities of a stochastically growing industrial park. By explicitly modeling randomness in the arrival of new tenants and their random wastewater discharges, our model calculates the future mean and variance of wastewater flow in the industrial park. Mean and variance are used in a Mixed Integer Programming Model to optimize wastewater treatment plant selection over a long planning horizon (30 years). By fitting our first model to empirical data from an industrial park in Turkey, we find that considering the variance of wastewater load is critical for long-term planning. Also, we quantify the economic significance of lowering wastewater discharges which can be achieved by water recycling or interplant water exchange. © 2023 Elsevier B.V.
Real Time Prediction of Delivery Delay With Machine Learning
(Kadir Has Üniversitesi, 2023) Küp, Büşra Ülkü; Hekimoğlu, Mustafa
İnternetin yaygınlaşması, e-ticaret ve lojistik endüstrilerinde önemli bir dönüşüme yol açmıştır. Bu dönüşüm, çevrimiçi alışverişte önemli bir artışa öncülük etmiş ve rekabetçi ortamda kargo şirketlerinin operasyonel verimliliğini arttırma ihtiyacını ortaya çıkarmıştır. Teslimat süreçlerini optimize etmek ve müşteri memnuniyetini artırmak amacıyla, makine öğrenimi kullanılarak teslimat gecikmelerinin tahmin edilmesi, lojistik şirketlerine önemli katkılar sağlayacaktır. Ayrıca, gerçek dünya verilerinin bu çalışmada kullanılması, elde edilen sonuçların güvenilirliğini artırmakta ve makine öğreniminin lojistik endüstrisi odaklı akademik araştırmalarda kullanılmasının avantajlarını vurgulamaktadır. Bu çalışmada, Logistic Regression, XGBoost, CatBoost ve Random Forest gibi en yaygın kullanılan dört denetimli sınıflandırma algoritması, bir e-ticaret lojistik şirketinde gerçek zamanlı veriler kullanılarak teslimat gecikmelerinin tahmin edilmesi amacıyla uygulanmıştır. Tüm süreç boyunca sürekli gecikme tahmini yapabilmek için, tüm teslimat süreci farklı gönderi türleri için sırasıyla 11 ve 15 adım şeklinde ayrıştırılmış ve her adım için ayrı tahmin modelleri oluşturulmuştur. Bu modellerin performansını artırmak için optimal parametre ve öznitelik seçimi yöntemleri kullanılmıştır. Kullanılan bu optimizasyon teknikleri, modellerin performansları üzerinde önemli bir olumlu etki sağlamıştır. Elde edilen sonuçlara göre, dört farklı sınıflandırıcı kullanılarak oluşturulan modellerin nihai ROC-AUC skoru ile değerlendirildi. XGBoost için ROC-AUC puanları \%71,5 ile \%99,9 arasında değişmekteyken, CatBoost için ROC-AUC puanları \%72,4 ile \%99,9 arasında değişim gösterdi. Bu iki sınıflandırıcı farklı adımlarda çok yakın performans göstermiş olsalar da, CatBoost genel olarak XGBoost'a kıyasla biraz daha iyi bir sonuç ortaya koymuştur. Gelecekteki çalışmalarda, daha doğru sonuçlar elde edebilmek için derin öğrenme bazlı sınıflandırma methodlarının denenmesi ve ek özniteliklerin entegre edilmesi üzerine çalışmalar yapılacaktır. Daha büyük veri kümeleri kullanılması önerilen gecikme tahmini yaklaşımının, daha etkin çıktılar ve performans iyileştirmeleri sağlayacaktır. Ancak, daha büyük veri kümeleri elde edilmesi, işlenmesi ve derin öğrenme modellerinin denenmesi için daha yüksek performanslı donanımsal, işlemci ve hafıza, kaynaklara ihtiyaç duyulacaktır. Bu zorlukların üstesinden gelmek ve daha yüksek performanslı çözümler sunmak için çeşitli stratejiler ve teknikler geliştirilmeye devam edilecektir.