Theses supervised by Dr. Öğr. Üyesi Erinç Albey
11 theses · Özyeğin University
A decomposition-based metaheuristic approach for solving the rapid needs assessment routing problem
This study proposes a decomposition-based tabu search algorithm for a multi-cover routing problem (MCRP), which aims to classify and evaluate the impacts of the disaster in different sites and the needs of different community groups affected by a disaster when remote communication is not possible, and highlights the solution time and quality performances of the introduced algorithm. The algorithm focuses on decomposing the problem to three phases and to apply different methods while solving them. Performance of the proposed tabu search algorithm is evaluated with respect to different benchmark solutions, findings are put to statistical tests, and the results indicate that the proposed algorithm can achieve high-quality solutions expeditiously, providing better results on average compared with the best-known solutions existing in the literature.
Makine öğrenimi görevleri için modern veri yönetim stratejileri: Bulut platformu üzerinde spor analitiği uygulaması
There is no doubt that data is the most valuable asset today. The efforts of enterprises in digital transformation and creating a data-driven culture are the most concrete indicators of this. Nowadays, where data transforms all industries, it is possible to follow the rapidly developing technological developments in this field. Appropriate data management strategies are the basis of creating data-driven organizations. When the evolution of data management architectures is examined, it is possible to say that the biggest factor that triggers this evolution is the changing and increasing data sources and the velocity of data production. In addition, with the increase in the importance of business use cases that need to be done in real time, it has become a very crucial need to process data quickly and turn it into action. Today when data is strategic importance, enterprises that can manage data correctly could gain competitive advantages. Being able to the correct data management can be built with the support of up-to-date and modern approaches. The infrastructures established by the integration of new and modern methods into the platforms turn into more agile structures. This increases the number of value added services to be produced from data by providing speed and flexibility to organizations. Today, the outputs expected to be produced from data management platforms go beyond descriptive and diagnostic analytic. Now, artificial intelligence, machine learning and data science are important parts of these platforms and these are opened new channels for the future of enterprises. In this thesis, basic needs and capabilities of modern data management architectures are described and detailed explanations were made on reference architectures in the industry. Besides, data management strategies and expectations were discussed. An example prototype of the data management platforms, which is explained in detail in this thesis, has also been developed on the cloud platforms. In this prototype, the entire life cycle of the data was considered and each step was developed in detail. In addition, a data science project was developed using the data collected on the platform. Thus, an end-to-end solution has been implemented.
Plastik enjeksiyon prosesinde eksik enjeksiyon hatasının engellenmesi
The Plastic injection molding machine (IMM) has been one of the most preferred manufacturing methods since its invention and continues its development in parallel with the advancing technology. The demand for faster, better quality, and cheaper parts with improved quality output is increasing day by day. Manufacturers are dealing with quality errors in production while trying to meet this demand. The most common of these quality errors are the short shot missing injection error. The special mold investment, changing of part design, and usage of distinct plastic raw materials are carried out to solve this error. The most practical temporary action taken to prevent this error in production is to increase the raw material's weight by optimizing the process parameters. Unfortunately, all of these solution methods increase the product cost. This thesis aims to develop a system for preventing short shot quality failure without using a solution to massive increase the part's cost. Firstly, the finite element analysis (FEA) for the short shot quality defect in injection machinery molding (IMM) is performed via Autodesk Moldflow Analysis. The mold flow FEA model is verified in serial production. Then, IMM is integrated into manufacturing execution system (MES) and a quality control system with cameras installed at the end of the production line. The relationship between short shot quality defect and IMM critical process parameters extracted with data analytic. The model is generated and commissioned in the production line. The integration of each production line into MES and installing a camera measurement system can adversely affect industrial deployment sustainability. Therefore, a different approach is developed with in-mold pressure/temperature sensors. The production parameters affecting the short shot quality failure are determined via the data analytics. The goal is to produce the system in the ideal range in each cycle by giving feedback to the IMM according to the in-mold sensors' data during production. The main aim of this thesis is to avoid short-shot quality failure in plastic part production. The short-shot quality defect has not been eliminated in this study. However, it has decreased from 3.7% error rates to 0.82%. This dramatic reduction results in an overall 80% improvement in short shot quality failure without increasing product cost.
Yonga levha tesisi için uygulama: Kalite tahminlemesi ve dijital dönüşüm için web tabanlı karar destek sistemi
As it is same for most of the production procedures, a certain quality level must be derived in the particle board production. In the particleboard production, a series of samples taken from the production line for the quality analysis of the products produced. These samples are put to the test for analysis and observe whether the quality metrics are satisfying the needs or not. Sampling can be done after a set of production is completed. Laboratory tests samples for at least three hours. If the results of the test are out of acceptable limits, then the facility changes the production parameters, waits for new production output to gain new samples, tests new samples again. While quality testing procedures are running, the products that do not satisfy quality limits, cannot be delivered to customers, which results in crucial capacity loss. In this study a decision support system is developed to measure real-time effects of changes of production parameters on quality metrics by using machine learning based prediction models with live production data collected from production line. Decision support system developed in this study enables to predict three different quality metrics while margin of error is realized around 5%, on the average.
Müşteri şikayetlerini otomatize edebilmek için uygulanabilir yaklaşım
Customer complaint management is critical and time-consuming process for institutions. For an effective management and increased customer satisfaction, developing an instant and automated reply mechanism is essential. This phenomenon leads the motivation which helps data scientists to work for developing chatbots. For complex chatbots state of the art techniques of Natural Language Processing is essential to catch intend whereas its high costs. On the other hand, basic machine learning algorithms which is enhanced with NLP techniques can be more applicable for limited data resources. Also even with the help of Regular Expression techniques, quick and effective solutions can be achieved. This thesis covers the development process of a primitive chatbot which is developed by using the customer complaints which are in Turkish language.
NFT pazarının tanımlayıcı ve tahmine dayalı analizi
Non-fungible tokens (NFTs) are digital assets on a blockchain that have unique identi- fication codes and metadata that make them distinguishable from one another. NFTs can represent a wide range of digital assets, including game cards, artwork, and even real estate. Due to these characteristics, NFTs have gained a tremendous interest from people around the world, leading to huge returns on investment in the NFT market. However, there are only a few studies on the market in the literature. This paper examines various aspects of the NFT market to shed light on its dy- namics and wallet behaviors. First, a descriptive analysis of the market is performed to show its overall trend. The transactional behaviors of wallets are then analyzed, and a segmentation is made to gain a general understanding of the user portfolio. The buyers of a specific NFT collection (Bored Ape Yacht Club) are then studied by comparing them to the overall market, revealing differences in transactional tenden- cies and macro indicators. Finally, machine learning models are developed to predict the transactional behaviors of wallets. Our analysis has revealed that the growth of the NFT market is largely driven by new entrants to the market, but lately there has been a significant decrease in the number of new wallets entering the market. We have also found that the majority of wallets in the market have only one transaction and hold only one token, suggesting that these are users who are experimenting with the market. When we look at the Bored Ape Yacht Club sample, however, we see that these users are highly engaged with the market, with high trading frequencies and a diverse portfolio. Finally, our predictive models show that the transactional behaviors of wallets can be predicted, which opens up opportunities for optimization in various areas.
Resim işleme ile kripto varlıkların gelecekteki fiyat hareketinin tahmin edilmesi
Digital or virtual currency known as cryptocurrency uses cryptography for security and is not controlled by a central bank. Cryptocurrencies control the issue of new units and record transactions using decentralized technology, such as blockchain. Cryptocurrencies are entirely digital and have no physical form, in contrast to traditional currency, which is real and backed by a government or financial institution. Although Bitcoin was the first and best-known cryptocurrency, there are now thousands of other coins in use, including Ethereum, Tether, BNB, XRP etc. Bitcoin's value can be extremely unstable and it is frequently utilized as an investment or speculative asset. In some locations, it can also be used to make purchases of products and services, and some companies even accept it as payment. Due to the formation of this sector in relation to the growth in earnings and followers, all focus turned in the direction of the cryptocurrency market. Despite the abundance of studies that have been done in the past for this area, less image processing research has been done for the next movement prediction. This paper uses Bitcoin (BTC) dataset and tries to create a tool to project the upcoming price direction. The target variable is a binary type as the next movement will decrease or increase direction. The challenge of forecasting the next day's price movement involves learning from the information from the previous day by transforming them to the images. Since the main goal is using image processing for prediction, Convolutional Neural Networks, one of the most well-known deep learning techniques, and human judgment will be used. As a result, it is aimed to create an algorithm that makes a successful buy-sell decision on BTC and to achieve profitability. However, if the algorithm created for this study has adequate performance and structure on BTC, it will be simple to adapt it to other cryptocurrency kinds in the future.
A revised approach to cryptocurrency portfolio optimization using advanced Q-learning and policy iteration frameworks
Despite all the factors that cause concern among investors, such as volatility and decentralization of crypto world, the popularity of cryptocurrencies continues to grow steadily. The cryptocurrency market still holds its allure for many investors due to the high profit levels it has experienced in the past. With the entrance of numerous altcoins into the market, portfolio management becomes much more challenging. In the literature, we come across numerous studies proposing efficient portfolio management techniques for cryptocurrencies. This study presents proposed models developed based on policy iteration and Q-learning algorithms. Under Q-learning, three distinct sub-models are introduced: Deep Q-Network (DQN), Double Deep Q-Network (DDQN), and Double Dueling Q-Network (DDDQN). All of these models are trained using 6-month training periods and compared using 10 different training and testing periods. Additionally, to evaluate both of proposed policy iteration and Q-learning models, baseline models were created for each algorithm, and the performance of the proposed models was assessed against these baseline models. The results indicate that among Policy Iteration models, the proposed model has the highest average ROI value of 3%, making it the top-performing model. Similarly, among Q-learning models, the proposed DQN model surpasses both baseline models and other Q-learning models, with an average ROI value of 2%. Considering all the models, the proposed Policy Iteration model achieves the highest average ROI value, while the proposed DQN and the proposed DDDQN model demonstrates the lowest volatility in terms of ROI standard deviations.
Kanban ile çalışan yazılım geliştirme ekiplerinin kaynak optimizasyonu
Kanban has been a commonly used agile and lean way of working for software development teams due to its evolutionary approach, effectiveness in project management, and wide range of applicability. This thesis takes a holistic approach to investigate the optimization of software development teams working with Kanban. Due to the complexity of the nature of work, which involves multiple workstations, resources with different skill sets, work types, blocking scenarios, and work-in-process limits, simulations are conducted to gather data. The study proposes a solution framework that approaches the problem at two levels and an additional exploratory multi-objective optimization section using modern techniques to delve deeper into potential directions. The first level of the solution framework starts with a novel, comprehensive simulation algorithm that introduces a blocking scenario applied to the work items, unlike the machine breakdown method used in the manufacturing processes. The simulation model takes input parameters derived from a real-life company, adds resources, and WIP limits as other input parameters and collects data including output, lead time, blocking time, efficiency of resources, and more. This simulation module is then plugged into the solution methods including a greedy heuristic, a two-step model, and a decision tree clearing function to compare the three methods and find the best-performing under different conditions. The second level of the solution framework extends the simulation model by adding resource capabilities and work type priorities ending in over tens of millions of potential combinations. Due to the intractability of generating all possible combinations, three machine learning algorithms -random forest, XGBoost, and neural networks- are used to find the best potential feature set. Finally, a holistic approach is taken to account for lead time, output, and cost via multi-objective optimization models. Numerical analysis provides insights into the impact of resources, and work-in-progress limits through three strong evolutionary algorithms, which are strength Pareto evolutionary algorithm-II, Pareto archived evolution strategy, and non-dominated sorting genetic algorithm-II. The application of business priorities and the effect of these choices are demonstrated in navigating the trade-offs between cost, lead time, and output. This dissertation contributes significantly to both academic knowledge and practical applications, presenting pioneering insights into optimizing the output and lead time of a Kanban software development team, impacts of resources, work-in-progress limits, and priorities and provides a set of state-of-the-art practical solutions for achieving this goal.
Makine öğrenmesi tabanlı optimizasyon: Tutundurma yönetimi ve sıralama toplulaştırma uygulamaları
This thesis examines the synergy between Machine Learning (ML) and Operations Research (OR) to tackle complex optimization challenges in digital marketing (through proposing a genuine retention management framework) and sharing economy (via introducing a rank aggregation based matching approach for ride pooling problem). It aims to develop innovative models that combine ML's predictive accuracy with OR's decision-making strategies to enhance customer retention and optimize ride-sharing operations. The research follows a structured progression from theoretical concepts to practical applications, leading to creation of sophisticated algorithms that demonstrate significant improvements in operational efficiency. The introduction outlines the synergistic potential between ML and OR, emphasizing the need for a cross-disciplinary approach to solve real-world problems. It focuses on the application areas of customer retention and ride pooling. The reasoning behind these application areas is that these fields posses digital data, which makes possible to deploy advanced analytical techniques. Developed applications reaffirms the successful integration of ML and OR, as evidenced by enhanced decision-making processes and operational efficiencies in targeted applications. The results show notable advancements in predictive models for customer retention and a data-driven ride-matching algorithm, both of which significantly outperform the existing methods. Future research is directed towards developing more advanced algorithms that can handle larger datasets and deliver real-time analytics. The thesis also encourages further interdisciplinary studies to explore complex systemic challenges, promising to extend the scope and efficacy of the proposed models.
Tahmin sonrası optimizasyon ve kısıt öğrenme: E-ticaret karar alma süreçleri için ampirik bir karşılaştırma
In today's rapidly evolving e-commerce landscape, effectively balancing price and commission strategies is vital for platforms aiming to maximize both merchant profitability and their own revenue. This MSc. thesis introduces an innovative approach by integrating advanced machine learning techniques with optimization frameworks, specifically comparing the traditional Predict-then-Optimize (PTO) method against the integrated Constraint Learning (CL) paradigm. Through rigorous simulation and empirical evaluation using real-world marketplace data, the proposed frameworks demonstrate clear benefits in strategic decision-making, efficiency, and robustness. Ultimately, this research paves the way for smarter, more dynamic pricing decisions in complex online retail ecosystems.