4
Archived Theses
0
DOIs Assigned
0%
DOI Rate
Discipline
Bulanık mantık ile reklam kampanyaları için teklif optimizasyon modelinin geliştirilmesi
Turizm sektöründeki rekabetin internet ortamına taşınması ile önce internet ortamından rezervasyon yapılan online seyahat acenteleri (OTA) ardından da birçok otel meta arama motoru (metasearch) geliştirildi. Günümüzde kullanıcılar, bu metasearch'ler sayesinde bir otelin birçok farklı OTA'daki fiyatlarını karşılaştırabilir hatta bu metasearch'ten ilgili linke tıklayarak rezervasyon yapabilir. Bu çalışmada; Türkiye'nin önde gelen OTA'larından Tatil sepeti'nin, dünyanın önde gelen metasearch'lerinden Trivagodaki kampanyalarına, bulanık mantık modeli ile reklam teklifi önerisi verdik. Tatilsepeti'nin kampanyalarının dönüşüm miktarını ve görünürlüğünü maksimize etmeyi, tıklama başına maliyeti (CPC) optimum bir pozisyonda tutmayı amaçladık. Kullanıcı kısıtı ve maliyet kısıtı ile kampanyaların yönetilmesini daha güvenli hale getirdik. Seçilen kampanyaların gelir, maliyet ve tıklama sayıları tarihsel olarak karşılaştırıldığında optimizasyon modelimizin başarılı sonuçlar verdiği tespit edilmiştir.
Integrating fuzzy logic into portfolio optimization: Enhancing risk management and return maximization
Modern Portfolio Theory (MPT), or mean-variance analysis, is built on principles of risk and return. A core assumption in MPT is investor risk aversion, which significantly influences investment choices. In evidence, highly risk-averse investors often choose safer investments with lower returns over riskier options that may yield higher returns. This study seeks to optimize investment portfolio utility scores by modelling risk aversion using fuzzy logic. Introduced by Lotfi Zadeh, fuzzy logic addresses human reasoning and decision-making under uncertainty and ambiguity. Thus, this logic is suitable to model investors' diverse utility preferences and to reflect their subjective and varied perceptions of risk. Traditional utility scoring uses fixed risk aversion coefficients from 1 to 5, with higher values indicating greater risk aversion. Here, we propose a more granular approach, measuring risk aversion on a 0-100 scale, where 0 indicates total risk tolerance and 100 signifies complete risk aversion. We categorize investors as low (5-45), moderate (45-75), or high (75-95) risk-averse, acknowledging that no investor is entirely risk-seeking or risk-averse. Each risk aversion category is modelled with Trapezoidal and Cauchy distributions using a linear combination of their respective fuzzy membership functions. A sensitivity analysis was conducted, over short, mid and long-term timeframes, on a portfolio consisting of 10 assets which is optimized with fuzzy risk aversion coefficient then compared to the same portfolio optimized with fixed coefficients. Findings showed that fuzzy logic consistently provided superior risk-adjusted returns, particularly in mid-term and long-term scenarios. Although fuzzy-optimized portfolios had higher betas and variances, those risks were offset by enhanced returns. The Treynor ratios were similar across both methods, yet the fuzzy approach delivered higher overall returns, highlighting its effectiveness in managing volatility and risk.
BTK ve dijital platformlar üzerinden telekomünikasyon sektöründeki müşteri şikayetlerinin analizi
In this study, customer complaints shared on digital platforms were systematically analyzed. Data were collected from sources such as Şikayetvar, Google Play, and the App Store and processed using natural language processing (NLP) techniques. The primary objective was to identify the main problem areas in the telecommunications sector by applying sentiment analysis, topic modeling, and keyword extraction methods to the complaints. The data preprocessing phase involved cleaning the text data, handling missing values, and removing irrelevant content to ensure reliable analysis. Sentiment analysis was used to assess customer satisfaction levels, while topic modeling techniques helped uncover the most frequently mentioned issues. In addition to digital platforms, official complaint data obtained from the Information and Communication Technologies Authority (BTK) were also examined. A comparative analysis was conducted between digital complaints and official complaints submitted to BTK, focusing on factors such as complaint themes, resolution times, and customer satisfaction rates. The comparison revealed that complaints shared on digital platforms tend to spread more rapidly and are often less structured, whereas BTK complaints are more formal, detailed, and processed through standardized procedures. These findings highlight the complementary roles of digital platforms and official regulatory channels in capturing customer feedback. The study demonstrates that integrating insights from both sources can contribute to more effective customer service strategies and informed decision-making in the telecommunications sector. Keywords: Telecommunication, Complaint Analysis, Natural Language Processing, Customer Satisfaction, Digital Transformation
Tekstil sektöründe perakende satış analizi ve tahminlemesi
This thesis addresses sales forecasting in the retail domain using machine learning methods. Nowadays, the retail sector emerges as a rapidly changing and exponentially growing field. Therefore, obtaining accurate and reliable predictions in sales forecasting holds critical importance for businesses to sustain their competitive advantage. Machine learning is recognized as an effective tool with data analysis and pattern recognition capabilities to address complex problems like sales forecasting. This study aims to tackle the sales forecasting problem in the retail sector and investigate how machine learning methods can be utilized to solve this issue. Firstly, an exploration of the existing sales forecasting methods in the literature and machine learning algorithms is conducted. Subsequently, the effectiveness and performance of various machine learning algorithms (such as LGBM, LSTM, XGBoost) in sales forecasting are compared. This thesis is supported by experimental studies conducted on real-world datasets. The datasets encompass sales data from the textile retail sector and comprise data obtained over a specific time frame. Analyses conducted on these datasets reveal how machine learning algorithms can offer an advantage in sales forecasting compared to traditional methods. The results demonstrate that machine learning algorithms constitute an effective and accurate tool for sales forecasting in the retail sector. This thesis provides valuable insights into the LightGBM method specifically chosen and can aid businesses in enhancing strategic decisions, such as demand management, inventory optimization, and stock planning.