Prof. Dr. Ceyda Oğuz danışmanlığındaki tezler
16 tez · Koç University
Tramp shipment routing and scheduling with inventory and stowage constraints
Maritime transport has been a fundamental mode of transportation for centuries, enabling the transportation of heavy and bulky items in large quantities across oceans and seas. The use of ships has allowed for the efficient movement of goods and commodities and has been vital in facilitating global trade and commerce. In modern times, maritime transport has become even more critical with the exponential growth in global trade and the movement of goods, making it an essential component of the global economy. There are various types of operations involved in maritime transport, with liner, tramp, and industrial shipping being the three most commonly used. Liner shipping refers to the use of regularly scheduled services between designated ports, where ships follow a fixed timetable and transport a specific type of cargo. In contrast, tramp shipping operates on an ad hoc basis, with ships not following a fixed schedule but instead sailing according to the requirements of customers, and the ships are usually rented or leased. Industrial shipping, on the other hand, refers to the transportation of goods using ships that are owned by the supplier or customer. Tramp shipping alongside with the liner shipping is widely used in maritime transport, particularly for the transportation of bulk cargoes such as coal, iron, and grain. In this type of shipping, ships are chartered on a single voyage basis, and the cargo is loaded and transported according to the specific requirements of the customer. The ships used in tramp shipping are usually equipped with cranes to load and unload the cargo. In the tramp shipping problem considered in this study, the objective is to transport a set of cargoes from their origin ports to their respective destination ports using a homogeneous fleet of ships. The origin and destination ports of both the cargoes and the ships are predetermined. All of the cargoes are mandatory and have to be sent to their destination point within their respective deadline. If a cargo arrives at its destination point before the deadline, there will be a deduction from the cost as a reward. To minimize the sailing cost, inventory cost, and stowage cost, the problem is modeled as a mixed integer linear program, taking into account the stowage and inventory constraints that arise in real life applications. Stowage constraints are used to ensure the stability of the ship and to eliminate the loss of time when loading and unloading the cargoes. The ports have initial inventories of the cargoes, and their inventory value changes according to the amount shipped. Inventory constraints are developed to control the changes in inventory levels. Decisions must be made based on the routes and schedules of the ships, as well as the stowage and inventory constraints. The computational results obtained from 153 instances provides valuable insights for solving the tramp-shipping problem. The results show that while small-sized instances can be solved optimally, medium-sized and large-sized instances achieve gaps between 5-30% and 30-70%, respectively, depending on different problem characteristics. This gap is found by comparing the current solution to the best bound found in the problem. This study of the tramp shipping problem is designed to provide significant practical applications in the maritime transportation industry, as it can help shipping companies optimize their fleet utilization and reduce their transportation costs, leading to increased efficiency and profitability. Previous literature concerning the combination of these three maritime problems was limited, this study is designed to fill a significant gap in the literature and bring valuable insights for further research.
Scheduling of resource constrained projects via genetic algorithm with multiple skills of resources in multi-project environment
This thesis presents a genetic algorithm with invasive weed optimization reproduction for the multi-skill resource constrained multi-project scheduling problem with global and local resources (MS-RCMPSP). The problem is constructed by considering skill and skill levels of the resources and geographical competencies of them. The precedence relationships among tasks belonging to multiple projects is also considering, when scheduling the projects. The proposed method is tested with 30 generated instances and these instances are categorized as small, medium, and large. The performance of the algorithm and the parameter selections are separated into small, medium, and large categories, and performance evaluations are made by comparing the standard genetic algorithm and invasive weed optimization. The effectiveness of the proposed algorithm has been validated by the computational results and the algorithm's performance in this thesis is compared with the standard genetic algorithm (GA) and invasive weed optimization (IWO).
The empirical analysis of international climate policies for energy decisions
While the Paris Agreement enforces a radical transformation of energy systems to limit global temperature rise to below 1.5°C, market-based economic instruments, like the European Union Emissions Trading System (ETS) and the Carbon Border Adjustment Mechanism (CBAM), accelerates the internalization of carbon emission costs internationally. This pressure demands transparent, feasible, and comparable planning tools to safeguard supply security while controlling costs. However, significant gaps persist: data-intensive general equilibrium models fall short in responsiveness, and machine learning-based prediction models, as "black boxes", lack sufficient auditability and transparency for policymakers. This dissertation, comprising a complementary series of four papers, aims to fill the identified gap by developing a multi-layered hybrid methodology building on the global energy analysis covering 147 countries and extending it to national dynamics of Türkiye and electricity sector in detail. This study integrates computationally efficient partial equilibrium economics, multi-output machine learning, and optimization methods to provide a controllable trade-off between cost, carbon, and energy security under ETS/CBAM conditions across various carbon price scenarios. It proposes an innovative planning perspective that simultaneously tackles the carbon constraints of the Paris Agreement along with the goals of energy security and fiscal efficiency, through four consecutively structured papers spanning global, national, and sectoral scales. In the first paper, the partial equilibrium model replicates the fossil fuel share of 147 countries from 1997–2014 with an average absolute error of 0.28–0.34%, demonstrating that carbon tax-incentive packages align with the 2°C target without disrupting budget balance. The second paper models the governance update period (GUP) using multilevel machine learning, quantitatively determining an optimal policy cycle that further reduces the fossil share by 6 percentage points. The third paper, scaled to the national level, assesses Türkiye's electricity installed capacity for 2030 by comparing a parametric depreciation model with LSTM/GRU-based deep learning estimates, where the deep learning approach significantly reduces the normalized error rate in hydropower generation. The Green Growth scenario increases the renewable energy share from 56% to 79% while lowering carbon intensity to 160 kg CO₂/MWh. In the fourth paper, multiple output (MO) predictions are integrated with set-covering and two-stage grid-search optimization, reducing the total root mean square error (RMSE) to 0.587, boosting renewable energy capacity by 2.2 times compared to 2024, and enabling the grid-search method to achieve the lowest CO₂ emission profile. The collective findings from these four papers underscore the development of a multi-layered hybrid methodology that delivers three unique contributions: (i) it facilitates seamless integration of global modeling outputs with low computational overhead into national-level machine learning models; (ii) it directly models simultaneous relationships among diverse energy sources through its multi-output architecture; (iii) it employs optimization tools in bidirectional interaction with machine learning forecasts, enabling a holistic evaluation of decision-making, cost, and risk dimensions within a unified framework. Findings indicate that robust carbon pricing and data-driven capacity allocation, paired with a long-term planning perspective, can align carbon intensity with Paris Agreement targets while minimizing total system costs. Consequently, methodology developed in this dissertation equips policymakers with a scalable, auditable, and feasible decision-support tool within the ETS–CBAM environment.
İhracat konteynerlerinin elleçleme sayısını enazlamak için sezgisel yöntemler
In the growing international trade, the number of import, export and transit containers is ever increasing. Therefore, the importance of the container terminals and their efficient managements are highlighted. In this thesis, we study stacking policies in a container terminal for export containers, due to their characteristics. Different than the literature, we considered both the space allocations for containers arrived recently and the pick up operations before the departure of the containers. We assume that an initial configuration exists in the storage yards for already stored containers. The containers arrived recently are allocated to the blocks in the storage yard based on this initial configuration by taking the departure time of the containers into account, which are known beforehand since they are export containers. Hence the retrieval sequences of the containers within blocks are known. Another important aspect of our study is the inclusion of the remarshaling operation, which is used to speed up the retrieval of the export containers. Throughout these stacking operations (allocation, remarshaling, and pick up), some containers already stored in the blocks might be moved into other positions and these moves are known as relocations. These relocations cause major time and cost expenses in the container terminals since any relocation may result in several rehandling operations. Hence the main focus of this thesis is to deal with the rehandlings. We propose several heuristic approaches to estimate the locations for the containers arrived recently and maybe relocated export containers in order to minimize the number of rehandlings. We then analyze the performance of these proposed heuristic algorithms with a set of randomly generated problem instances considering different initial configurations under different container terminal scenarios.
Rıhtım vinci çizelgeleme problemi için kısıt programlama yaklaşımı
Minimizing the average vessel berthing time is one of the challenges for container terminals. Since containers are deployed from vessels by a quay crane, operations of this huge equipment may cause a bottleneck for the overall performance of a terminal. This study examines the quay crane scheduling problem (QCSP) at the seaside of container terminals. The QCSP requires completion of all loading and unloading operations of a berthed vessel. A constraint programming (CP) model, which consists of global constraints and propositional logic, is constructed by taking numerous properties of the problem such as safety margins, travel times and precedence relations into account. The performance of the proposed CP model is compared with algorithms presented in recent QCSP literature. The result from the computational experiments indicates that the proposed CP model is able to produce good results for the QCSP while reducing the computational time. Lastly, to show the robustness and the flexibility of the proposed model, extensions of the problem with ready times and time windows are also discussed.
Sipariş kabul etme ve çizelgeleme problemi için değişken komşuluklu arama
Order acceptance is one of the important decisions to make while dealing with satisfaction of customers, risk of delays and overloaded production in competitive environments. A company can increase its pro fit, satisfy demands of the customers and utilize its capacity at its best with a proper management of the incoming orders through making acceptance-rejection decisions on the orders and simultaneously scheduling the accepted orders. This problem is known as the order acceptance and scheduling (OAS) problem. In this study, we examine two diff erent OAS problems on a single machine environment. In the first problem, each order is characterized with a processing time, a due date, a weight and a revenue. Each accepted order which is delivered to the customer before its due date brings maximum pro fit to the manufacturer. Late delivery of an order causes tardiness cost which decreases the profi t. The manufacturer can reject the order without a penalty cost. Sometimes customers may specify deadlines for their orders. Deadlines are the preferred latest time for the customers to accept the orders. If the completion time of an order exceeds the deadline, the customer refuses the order and does not pay for it. Moreover, some orders coming from the customers may be defi ned with release dates to be ready for the processing.In the firrst problem, we ignore sequence dependent setup times (preparation time necessary between two successive orders), deadlines and release dates. The second problem includes these properties. The objective function for both of the problems is to maximize total profit that is a function of total revenue and total tardiness. We propose a Variable Neighborhood Search (VNS), which is a metaheuristic solution approach, to solve this NP-hard problem. The VNS is developed by using two neighborhood structures with a local search in a compact form. We analyze the performance of the VNS for both of the problems by using a benchmark data set. We present the computational experiments in which the VNS is compared with the most competitive metaheuristic algorithms from the literature. We conclude with the insights gained regarding the strengths and weaknesses of the proposed algorithm and that of the algorithms from the literature.
Çok ürünlü akış tipi üretim sistemlerinde kafile bölme ve kaydırma problemine sezgisel yaklaşımlar
In thesis research, we consider the multi-product lot streaming (MPLS) problem with equal and consistent sublots in multi-machine flow shops (MMFS) with objective of minimizing the makespan. We firstly introduce various types of scheduling problems and provide detailed background information and literature review on lot streaming problems with equal and consistent sublots. Then we develop two heuristic procedures for equal and consistent sublot sized MPLS problems respectively. First heuristic approach that we develop for MPLS problem with equal sublots in MMFS (heuristic RO) is a constructive procedure, which has many distinguishing characteristics. Its fast and easy construction method of initial lot sequence lets tabu search algorithm start with a better initial solution in contrast with random initial solutions. Moreover, we utilize the concept of ?Interior Lots? in order to restrict the insertion of a given lot into first position. We also provide a proof of the claim, which supports the use of ?Interior Lot? concept in both of the heuristics. Second heuristic approach (heuristic RO-C) deals with the MPLS problem with consistent sublots in MMFS. In compliance with the different characteristics of the problem we develop an additional tabu search algorithm in order to generate better initial sublot size matrix. Finally we present comparative results of experimental studies for both heuristics. We show that solution qualities of both heuristics that we develop are better than or equal to those obtained by the heuristic and exact methods that we choose to compare with. Proposed heuristics have considerable contributions to MPLS literature due to their unique ordering and tie breaking rules for sorting bottleneck dominant and reversely bottleneck dominant lots, utilization of interior lot concept in several steps of heuristics and solution quality with respect to similar studies from the literature. Keywords: Lot Streaming, Equal Sublot, Consistent Sublot, Bottleneck Dominance, Interior Lots, Tabu Search, Run-in Time, Run-out Time
Düzgeçiş konteyner terminalleri
The ever increasing number of containers to be transported necessitates expanding the capacity of container terminals. This can be achieved either by enlarging the container terminals physically or by managing the container terminals better. Since the first alternative is costly, and sometimes not possible due to limited land, container terminal managers focus on the second alternative. One possible way to increase the capacity of a container terminal is to reduce the vessel handling times. By that way, containers will spend little or no time at the yard storage phase, which will directly enhance the capability of container terminals to compete with each other. This setting is applicable to large transhipment hubs where many vessels can be moored and processed at the same time and discharged containers can directly go to their outbound vessels. Since the yard is a critical resource for the container terminal management, this environment will result in less cost and better utilization of the yard space. In this study, we consider a transshipment container terminal where all containers must be loaded to their outbound vessels directly after being unloaded from their incoming vessels. Our aim is to determine a schedule of unloading and loading times of containers while minimizing the makespan, which is the total completion time of all operations. This problem has not been studied before. The most related work deals with a case where there are two vessels and three types of containers, which are loaded, unloaded and directly transshipped ones. In our study, we consider just the transshipment containers but the number of vessels is not restricted to two. We develop a mathematical model to represent the system and analyse its performance through computational studies. We create instances from different sizes to understand the limits of the mathematical model. After identifying these limits for the model, we study possible solution methodologies.
Erteleme kısıtlı tek makine çizelgeleme
In a single machine problem a set of tasks has to be processed without preemption on a machine in such a way that at most one task is processed at any time and a given objective function is optimized. This thesis examines the single machine scheduling problems with timelag constraints. In these problems, precedence relations exist among the tasks that are to be scheduled. In a precedence relation, successor task is not allowed to start before the processing of the predecessor task is completed. Precedence relations may be generalized by adding the timelag constraints. In this thesis, timelags are assumed to be minimal timelags. Timelag constraint necessitates delaying the start time of the successor task after the completion time of the preceding task. Precedence constraints between tasks that have to be respected in every feasible schedule generally increase the computational complexity of a scheduling problem. Occasionally, their introduction may turn a problem that is solvable within polynomial time into an NP-complete one (Lenstra and Rinnoy Kan, 1978). Generalizing the problems with precedence constraints by introducing the timelag constraint makes problems harder. In addition to precedence and timelag constraints, the release time constraint is considered in this thesis. Even though there exists a wide variety of results on the complexity of scheduling problems with precedence and timelag constraints, there are still some problems for which the complexity status remains open. In this thesis, two scheduling problems with open complexity status are proven to be polynomially solvable. Also, a scheduling problem which is the generalization of two strongly NP-hard problems is considered and a branch and bound algorithm is proposed. Furthermore, for each problem, integer programming model is given. Finally, computational results are provided.
Sipariş kabul ve çizelgeleme problemleri için zaman endeksli matematiksel modeller
Scheduling has been an active area of research for decades. A substantial amount of work has been done to define, classify, and solve scheduling problems. Most of these problems are computationally di ffcult to solve, and incorporating other decisions into them increases the complexity of these problems. This thesis focuses on order acceptance and scheduling (OAS) problems, and proposes time-indexed mixed integer and linear programming (MILP) models for the fi rst time. An OAS problem can be defi ned with parameters of processing time, release date, due date, and deadline for each order. There are also a maximum revenue that will be brought by each order and an importance weight of each order. Furthermore, there could be a setup time between orders if they are processed consecutively. In the thesis, both the general OAS problem (OAS 2) that will be defi ned with all above parameters, and a special case (OAS 1) that will exclude the release dates and setup times are considered. After developing a time-indexed MILP for both OAS 1 and OAS 2, their efficiency and eff ectiveness were tested computationally. To enhance both the effi ciency and the eff ectiveness of OAS 1, three dominance properties were suggested, and it was observed that while the optimality gap was decreased, the computational time was also reduced considerably so that large instances can be solved. Since OAS 2 is more challenging than OAS 1, the time-indexed MILP developed could not solve problem instances to optimality in a reasonable time. Hence, diff erent methods were proposed to find near optimal lower bounds and upper bounds for the problem. To obtain good upper bounds, a Lagrangian Relaxation method as well as a linear programming (LP) relaxation method with three new valid inequalities were proposed, and it is shown that Lagrangian Relaxation finds only loose upper bounds, but LP relaxation with valid inequalities improves almost all upper bounds for all sizes of instances. To obtain lower bounds, a simple heuristic method and a revised MILP formulation were presented, and then, it is shown that they are able to solve only small instances to optimality.
Bir dagıtım ve toplama metodu ve kamyon rota çizelgelemesine uygulaması
Vehicle Routing Problem can be briefly explained as distributing goods to a set of customers by using set of vehicles which perform their movements by using appropriate transportation mode such as aviation, land transport such as rail and road, ship transport, pipeline and cable. In particular, the solution of a Vehicle Routing Problem requires determination of a set of routes, each performed by a single vehicle that starts and ends at its own depot, such that all the demand of the customers are fulfilled, all the operational constraints are satisfied, and the overall transportation cost is minimized. The Classical Vehicle Routing Problem (VRP) is one of the most popular problems in combinatorial optimization and it has allowed considerable applications to exact and heuristic solution studies. In this study, a real world application for Vehicle Routing Problem for Turkish Cargo in Balkan cities is carried out. Delivery routes and pickup routes are modeled and exact solutions are found in General Algebraic Modeling System (GAMS) for a given time period. For the delivery of cargos in Balkan cities a Heterogeneous Fleet of Open Vehicle Routing Problem Model is developed and for the pick up case of cargos a Heterogeneous Fleet of Multi-Depot Open Vehicle Routing Problem Model is developed. A combined backhaul model Heterogeneous Fleet of Vehicle Routing Problem Backhauls with model is written. Comparisons of these models are done. Turkish Cargo requested a GAMS-Excel interface, in order to use the outputs of our models. In GAMS-Excel Interface, a user-friendly application is developed for users of Turkish Cargo. Interface reads and writes the demand values in GAMS file for each month. With the GAMS-Excel interface users in Turkish Cargo can easily see the route for delivery and pickup models for the desired month with a simple notation.
Kafes model yapısındaki protein yapı tahmini problemini yöneylem araştırma bakış açısıyla ele alma
Protein structure prediction (PSP) consists of predicting the native structure of a protein from its sequence of amino acids by minimizing an energy function. The problem is of vital importance in medical science, molecular biology, biochemistry, and biophysics. PSP being NP-hard, even when abstracted to lattice models, is computationally challenging. In this study, we develop several mixed integer linear programming (MILP) models for PSP problem under lattices, along with variety of valid inequalities, and two symmetry breaking techniques. We then propose three metaheuristic algorithms. While the focus of our study is on the hydrophobic-polar (HP) model under cubic and square lattices, next, we address some of the drawbacks of the HP model, by proposing extensions of our optimization methods to other more sophisticated PSP models. Finally, we evaluate the performance of these optimization methods with computational experiments. We demonstrate that our MILP models outperform the state of the art models both in terms of running times in finding the optimal integer solutions, and in finding tight bounds on the objective value provided by linear programming (LP) relaxations of the models. We also show that the metaheuristic algorithms are able to find the optimal solutions for many benchmark instances. We then use the solution provided by these metaheuristic algorithms as initial solutions for the MILP models, and for constraining the feasible region. Integrated methods outperform significantly the state of the art models proposed for HP model in lattices. Finally, the computational results establish the robustness of our extended methods.
Kıyı terminali operasyonları için matematiksel modeller
Maritime terminals are the key components of global freight transportation as they handle over 80\% of global trade by volume and more than 70% of value according to United Nations. A steadily increasing workload causes maritime terminals around the world to face with a high competitive pressure. Therefore, it is essential for them to improve their performance levels in terms of service rate and costs. Maritime terminals can be classified into two based on the transported materials: container and bulk. Differently from container terminals in which standardized containers are processed, bulk terminals deal with unpackaged natural resources and agricultural products, such as iron ore, coal, grains, oil, and gas in large quantities. Operational problems observed in both classes of maritime terminals are the variants of well studied operations research problems such as machine scheduling, vehicle routing, and bin packing. However, they are more complex in nature because of the distinctive characteristics of terminals. In this dissertation, we study three different but related planning and scheduling problems of maritime terminals with a practical relevance: quay crane assignment problem (QCAP), reclaimer scheduling problem (RSP) and integrated dry bulk terminal (IDBT) problem. In each of these problems, we deal with one of the most challenging characteristics which is caused by the fact that bottleneck equipment in both classes of terminals, namely, quay cranes (QC) and reclaimers, are mounted on the same rail track, thus their movements are restricted by their respective positions over time. In the first chapter, we introduce container and bulk terminals by presenting a concise overview of their operations as well as the related literature. Within these sections, we also describe our motivations and contributions for QCAP, RSP, and IDBT problem, respectively. In Chapter 2, we study QCAP. In this problem, QCs are assigned to arriving vessels and handling time of a vessel depends on the number of assigned QCs. As QCs are the bottleneck equipment in container terminals, they need to be utilized efficiently. We study a QCAP by considering its various features, differently from the literature which deals with the simplest form of the problem. With our approach, we can estimate the vessel handling times accurately and hence improve the productivity as a result. We represent this problem as a moldable task scheduling problem with contiguous assignments, and we formulate an extended time-indexed model that additionally keeps specific task to machine information. Even though extended formulation is flexible in terms of modeling different features of the problem, it has multiple computational issues. Therefore, we develop an exact solution method by first hybridizing the formulation with a set of auxiliary variables to obtain a decomposable structure, and then implementing logic-based Benders decomposition (LBBD) with strong cuts. One limitation of this proposed method is excessive memory requirements for large instances, since it is based on a time-indexed formulation. Hence, we propose methods to derive lower and upper bounds for larger problem instances in Chapter 3. Since linear programming relaxations of time indexed formulations are known to be very tight, we developed a column generation procedure in which pricing problem can be solved in polynomial time. For an upper bound, we introduce a constraint programming model that finds near optimal solutions in a short time. In Chapter 4, we introduce the reclaimer scheduling problem. Reclaimers handle the dry bulk cargo, which is stacked as a stockpile. Reclaimers are mounted on the multiple rail tracks. If there are two reclaimers on the same rail track, then they cannot cross each other. Furthermore, a reclaimer can only handle stockpiles located adjacent to its rail track. For this strongly NP-hard problem, we opt for a heuristic approach by developing an arc-time-indexed lower bound model and constraint programming model to find near optimal solutions. There are many relations between operational problems in maritime terminals. If we solve these problems hierarchically, we often end up with plans with poor overall quality. Accordingly, in Chapter 5, we study the integrated problem dry bulk terminal operations, by considering berth allocation, yard assignment, and reclaimer scheduling operations simultaneously. After observing a key relation between problems, we decompose the problem into two easier problems and solve with a novel LBBD. In this decomposition, we model master and subproblems with mixed-integer programming and constraint programming, respectively, by exploiting the respective advantages of programming paradigms. Results show that the proposed method is able to solve considerably large instances to optimality in an acceptable time, compared to the monolithic approach. We conclude the dissertation with Chapter 6 in which we present a concise overview of our contributions and discuss future research directions.
Sipariş kabulü ve çizelgeleme problemleri için matematiksel modeller ve sezgisel algoritmalar
In make-to-order production systems, manufacturers often do not keep inventory of final goods beforehand which necessitates utilizing the production capacity more efficiently to be able to complete orders in more timely manner. Yet, it may not be possible to complete all orders on time when the order delivery time requirements of customers get more tight under a limited production capacity. In this case, manufacturers have to reject some orders and should simultaneously decide which orders to accept and how to schedule accepted orders since both decisions strictly depend on each other. The corresponding problem is called as the order acceptance and scheduling problem in the literature. In this thesis, we study the order acceptance and scheduling problem with release times and sequence dependent setup times that determines the set of accepted orders and their schedule so as to maximize the total revenue. In this thesis, a mixed integer and a constraint programming model, and a matheuristic algorithm along with a new relaxed time-indexed model formulation, a variable neighborhood and a tabu search algorithm to be used within are presented for the generalized order acceptance and scheduling problem. Computational results show that the proposed models achieve smaller optimality gaps than the existing models in the literature and the proposed matheuristic algorithm outperforms both the proposed models and the state-of-the-art algorithms for the order acceptance and scheduling problem in the literature. New optimal solutions are also identified by the proposed models and the matheuristic algorithm. Then, the order acceptance and scheduling problem is extended to include batch delivery of orders. In this extension, orders are not delivered individually (i.e., upon their completion) but rather are delivered in batches which is observed more often in real-life practices. Mathematical models developed for the original problem are adapted to this extension. To tackle large size problem instances in which these models do not perform well, we propose two different iterated local search (ILS) algorithms. The first ILS employs a variable neighborhood search algorithm with alternating objective functions and the second ILS utilizes a tabu search algorithm as its local search mechanism. Computational results show that the proposed models achieve small optimality gaps for the small size problems. However, their performances deteriorate significantly as problem size increases. For medium and large size instances, the first ILS algorithm using the proposed alternating objective functions achieves smaller optimality gaps than both the proposed models and the second ILS algorithm with tabu search.
Kanser hastalığında önemli gen kümelerini belirlemek için geliştirilen en iyileme modelleri
Using genomic characterizations of tumours biopsied from cancer patients has a great importance in understanding the formation and progression mechanisms in cancer. Survival analysis is one of the research methods that is used to predict overall survival time of cancer patients and to understand the aforementioned progression mechanisms. High dimensional structure of the genomic characterizations with the limited number of training samples makes survival analysis a challenging task. To be able to identify the survival associated biological mechanisms, cancer-specific pathway/gene set collections can be integrated into machine learning models. Existing approaches usually follow a two-stage approach that either identify predictive genes using a feature selection method and map these selected genes to known pathways/gene sets, or train separate models for each pathway/gene set and try to pick informative ones considering each model's predictive performance. Following such a two-stage approach might result in inefficacy of mapping selected genes to a known biological pathway/gene set due to highly correlated structure between feature groups or including related or very similar pathways/gene sets into the final model due to analyzing each pathway/gene set separately. In this thesis, rather than following such two-stage approaches, we propose machine learning models that can conjointly identify disease related biological mechanisms and perform survival prediction using only these identified biological mechanisms. Our algorithms obtain a sparse set of pathways/gene sets for the survival associated biological mechanisms by eliminating the uninformative ones from the model. We test our algorithms using 20 cancer datasets obtained from The Cancer Genome Atlas and two cancer-specific pathway/gene set collections as input data. We first propose a survival analysis model that integrates pathway/gene set collection into the model using multiple kernel learning. Our algorithm with conjoint modelling approach obtains statistically significantly better or comparable predictive performances against survival random forest (RF) and survival support vector machine (SVM) using significantly fewer gene expression features. Predictive performances of machine learning algorithms can be increased using multitask learning. For this purpose, we extend our multiple kernel learning-based algorithm towards multitask learning. Our multitask learning algorithm both models multiple cancer datasets simultaneously and integrates cancer related biological mechanisms into the machine learning model. The algorithm is able to identify common underlying biological mechanisms for cancer by obtaining better or comparable predictive results against survival RF, survival SVM, and our multiple kernel learning survival analysis algorithm. We also extend our multitask learning algorithm towards task clustering to identify the groups of cancer types that share similar underlying biological mechanisms. To this aim, we propose a unified formulation for task clustering, survival analysis, and knowledge extraction. Our clustering algorithm identifies relevant cancer groups by obtaining statistically significantly better or comparable predictive performances against survival RF, survival SVM, our multiple kernel learning and multitask multiple kernel learning survival analysis algorithms. Numbers of gene expression features and gene sets used by our clustering algorithm are significantly fewer than those of benchmark algorithms. These results show that our methods that identify survival associated biological mechanisms, obtain better or comparable predictive performances against survival analysis methods developed on genomic data without using the pathway/gene set information in the literature. In addition, we prove that survival prediction can be performed using fewer number of gene expression features compared to these benchmark algorithms. We also identify the cancer groups that share similar biological mechanisms without decreasing the predictive power.
Gıda bankası operasyonları için lojistik işbirliği
Supporting Sustainable Development is getting more and more important to leave a better world to the next generations and a sustainable food system is one of the elements in Sustainable Development. This study focuses on developing a sustainable food donation system operated for charitable agencies by an entrepreneurial company. Charitable agencies are mostly in the form of food banks. In the food donation system, donor companies donate their surplus food and the food banks are responsible to collect these donations assigned to them by a digital platform based on their demand. However, not every food bank has vehicles to collect donations on their own. In this study, it is proposed that the vehicles already owned by some of the food banks shall be used mutually to carry out the donation operations for other food banks too. This newly proposed concept is called logistics collaboration, and to optimize the system with collaboration, a mathematical formulation is developed. The optimization model assigns the donations timely to the food banks, arranges the daily routes of the vehicles, and decides the pickup and delivery operations to conduct throughout the routes. The model aims to maximize the total monetary savings of the donations rescued while minimizing the total environmental and economic damage caused by vehicle usage. To test the model, 136 real-life based instances are generated with varying sizes. Based on these instances, three case studies are created to evaluate the system with and without the collaboration. For the case studies, a GRASP-ILS metaheuristic implementation is also developed to improve the solution quality obtained by the model and to create starting solutions for the warm-up start procedures. It is shown that around 96% of the food donated can be rescued on average with the logistics collaboration. Furthermore, it is concluded that the model with collaboration outperforms the model without collaboration in terms of the objective function value of the best solutions and also in terms of the amount of rescued donations.