Browsing by Author "Sayar, Alperen"

Now showing 1 - 20 of 24

Attention-Enhanced Dual-Head LSTM With Rich Feature Engineering for Risk-Adjusted Stock Return Forecasting
(Institute of Electrical and Electronics Engineers Inc., 2025) Patel J.; Gunes P.; Ertugrul S.; Sayar A.; Benli H.; Makaroglu D.; Cakar T.; Benli, Harun; Gunes, Peri; Patel, Jay; Makaroglu, Didem; Sayar, Alperen; Cakar, Tuna; Ertugrul, Seyit
Stock return forecasting is a challenging task due to the complex, nonlinear, and volatile nature of financial markets. In this paper, we propose a comprehensive deep learning framework that integrates: a two-layer Long Short-Term Memory (LSTM) network augmented with a learnable attention mechanism, a dual-head output for simultaneous regression of next-day returns and classification of price direction, with an extensive suite of technical and macro-financial features. Our feature set comprises lagged log-returns, trend indicators (simple and exponential moving averages), momentum oscillators (RSI, MACD), volatility measures (rolling variance and GARCH conditional volatility), price bands (Bollinger Bands, Donchian channels), volume metrics (On-Balance Volume, Volume Rate of Change), Hidden Markov Model regime states, market index returns, and calendar effects. We train and validate the model using a rolling-window cross-validation scheme with early stopping and hyperparameter tuning to ensure temporal robustness. Empirical results on a large multi-stock dataset demonstrate that our attention-enhanced, dual-task LSTM outperforms single-task LSTMs and traditional machine learning benchmarks, achieving lower forecasting error and more stable generalization. © 2025 IEEE.
Customer Segmentation and Churn Prediction via Customer Metrics
(IEEE, 2022) Bozkan, Tunahan; Cakar, Tuna; Sayar, Alperen; Ertugrul, Seyit
In this study, it is aimed to predict whether customers operating in the factoring sector will continue to trade in the next three months after the last transaction date, using data-driven machine learning models, based on their past transaction movements and their risk, limit and company data. As a result of the models established, Loss Analysis (Churn) of two different customer groups (Real and Legal factory) was carried out. It was estimated by the XGBoost model with an F1 Score of 74% and 77%. Thanks to this modeling, it was aimed to increase the retention rate of customers through special promotions and campaigns to be made to these customer groups, together with the prediction of the customers who will leave. Thanks to the increase in retention rates, a direct contribution to the transaction volume on a company basis was ensured.
Citation - Scopus: 1
Distinguishing Cognitive Processes: a Machine Learning Approach To Decode Fnirs Data for Third-Party Punishment and Credit Decision-Making
(Ieee, 2024) Filiz, Gozde; Son, Semen; Sayar, Alperen; Ertugrul, Seyit; Sahin, Turkay; Akyurek, Guclu; Çakar, Tuna
Functional near-infrared spectroscopy (fNIRS) has seen increasingly widespread use in examining brain activity and cognitive processes. However, the existing literature provides insufficient information on distinguishing between different decision-making mechanisms. This study explores the application of fNIRS in differentiating between two distinct decision-making processes: third-party punishment decisions and credit decisions. The research includes analyzing fNIRS data collected during these processes and classifying the associated neural patterns using machine learning. The findings reveal that fNIRS, in conjunction with ML, holds substantial potential to enhance the depth of understanding of decision-making processes in neuroscience research.
Citation - Scopus: 3
Emg-Based Bci for Picar Mobilization
(IEEE, 2022) Yilmaz, Yasin; Günden, Burak Bahri; Ertekin, Efe; Sayar, Alperen; Çakar, Tuna; Arslan, Şefik Şuayb
In this study, the main scope was to develop a brain-computer interface (BCI) with the use of PiCar and EEG/ERP devices. Thus, it is aimed to facilitate the lives of people with certain diseases and disabilities. The ultimate goal of this project has been to direct and control a BCI-based PiCar concerning the signals captured via the EEG/ERP device. With the EEG headset, the EMG signals of the gestures (facial expressions) of the participant were captured. With the collected data, filtering and other preprocessing methods were applied to have noise-free signals. In the preprocessing, the detrending method was used to clean the data set which showed a constantly increasing trend, to a certain range, and zero trends. The denoising (Wavelet Denoising) and outlier detection/elimination methods (OneClassSVM) were used for noise elimination. The SMOTE oversampling method was used for data augmentation. Welch's method was used to get band powers from the signals. With the use of augmented data, several machine learning algorithms were applied such as Support Vector Machine, Logistic Regression, Linear Discriminant Analysis, Random forest Classifier, Gradient Boosting Classifier, Multinomial Naive Bayes, Decision tree, K-Nearest Neighbor, and voting classifier. The developed models were used to predict the direction that is passed as an input to PiCar's API. After that, PiCar was controlled concerning the predicted direction with HTTP GET requests. In this project, the OpenBCI headset and the Brainflow library for EEG/EMG signal obtaining and processing were used. Also, the Tkinter library was used for the Graphical user interface and Django for establishing a server on PiCar's brain which is RaspberryPi. © 2022 IEEE.
Emotion Recognition From Speech via the Use of Different Audio Features, Machine Learning and Deep Learning Algorithms
(AHFE International, 2023) Sayar, Alperen; Gümüş, Fatma; Çakar, Tuna; Ertuğrul, Seyit; Bozkan, Tunahan
Emotional Analysis of Candidates During Online Interviews
(AHFE International, 2023) Sayar, Alperen; Güvençli, Mert; Çakar, Tuna; Ertuğrul, Seyit; Bozkan, Tunahan
Citation - Scopus: 1
Enhancing Quality Control in Plastic Injection Production: Deep Learning-Based Detection and Classification of Defects
(IEEE, 2023) Mutlu, İsmail; Çakar, Tuna; Aslan, Yeşim; Yıldız, Ahmet; Sayar, Alperen; Şimsek, Kamil; Tunalı, Mustafa Mert
This study investigates the applicability of diverse deep learning techniques in detecting and classifying defects within plastic injection manufacturing processes. The findings derived from the models yield several feasible solutions that hold potential practical implications. Notably, the implementation of the Xception model as a classification framework presents a potential domain for enhancing quality control procedures. The developed models, trained on the prepared data sets, provide compelling evidence for the potential utilization of artificial intelligence technologies in the manufacturing industry. Consequently, this study represents a noteworthy contribution to the limited yet auspicious academic research in the field.
Fraud detection and prediction with machine learning applications
(MEF Üniversitesi, 2023) Sayar, Alperen; Çakar, Tuna
Bu çalışmanın temel amacı, faktoring sektöründe faaliyet gösteren bir şirketin müşterilerinin işlemleri üzerindeki dolandırıcılık faaliyetlerini tespit etmek ve buna bağlı olarak müşterilerin geçmiş işlem ve bağlantı verilerine dayalı keşifsel veri analizi ile ölçülebilir parametreler yakalamaktır. ve ardından hedef için tahmine dayalı modeller gerçekleştirmek. Sınıflandırma modeli algoritmaları olan XGBoost ve CATBoost modellerinde %79 civarında isabet oranı elde edilmiştir. Bu sayede dolandırıcılık yapma potansiyeli yüksek müşteri tespit edildikten sonra daha etkin, verimli ve doğru bir yaklaşımla hareket edilerek işlem bazında dolandırıcılık faaliyetlerinin doğrudan tespit edilmesi amaçlanmaktadır.
Citation - Scopus: 3
Grafraud: Fraud Detection Using Graph Databases and Neural Networks
(IEEE, 2023) Raina, Ajeet Singh; Çakar, Tuna; Ertuğrul, Seyit; Arslan, Şuayip; Sayar, Alperen
The issue of fraud has become a significant concern for many companies, particularly in the finance sector, but the traditional methods of detecting fraud are no longer adequate. Innovative technologies are necessary to identify complex fraudulent activities, and RedisGraph, a high-performance graph database, may offer a solution. With the assistance of neural networks, RedisGraph can accurately and efficiently detect fraudulent transactions in vast and intricate environments. Companies typically use a combination of Python and Oracle Databases to design fraud detection systems. which provide robust data management and real time AI processing capabilities. These technologies allow to create fraud detection systems that can determine fraudulent activities in real-time. But according to advancements of fraud methods only using of these systems not efficient nowadays. This article presents a proof of concept based on an essential use case of RedisGraph-powered neural networks in detecting financial fraud. It demonstrates the value of carefully employing Python and Oracle Database to construct and deploy real-time systems that can efficiently detect fraudulent activities.
Graph Theory-Based Fraud Detection in Banking Check Transactions
(Institute of Electrical and Electronics Engineers Inc., 2025) Behsi Z.; Memis E.C.; Ertugrul S.; Sayar A.; Gunes P.; Seydioglu S.; Cakar T.; Gunes, Peri; Memis, Emir Cetin; Sayar, Alperen; Cakar, Tuna; Ertugrul, Seyit; Seydioglu, Sarper; Behsi, Zeynep
Traditional banking fraud detection systems rely on rule-based approaches that analyze individual transactions in isolation, failing to capture complex relationship patterns indicative of coordinated fraud schemes such as check-kiting and artificial credit score manipulation. We p resent our study, a novel similarity-based graph theory approach that constructs weighted networks between check issuers using Jaccard Similarity Index and employs advanced graph analysis to identify suspicious entity clusters without requiring complete transaction relationship data. Our approach combines Jaccard Similarity Index for behavioral pattern analysis (addressing payee information unavailability) with comprehensive graph analysis including centrality measures, community detection, and anomaly identification. Through comprehensive evaluation on real banking data containing 458,399 transactions from 121,647 unique issuers - the largest confirmed dataset in fraud detection literature - we demonstrate the effectiveness of our methodology. Following parameter optimization using grid search methodology (similarity threshold: 0.55, risk percentile: 0.75), our study achieves competitive detection rates in optimal configurations with an average F1-score of 0.447 (±0.164) and peak performance reaching an F1-score of 0.557, while providing superior network topology analysis with 0.923 clustering coefficient. The system operates under significant data privacy constraints, lacking personal identification information (names, account numbers, IDs) and complete payee data. Despite these limitations, our study outperforms traditional approaches by leveraging similarity-based indirect relationships, and we project that performance could reach 85-95% levels with complete data access. © 2025 IEEE.
Citation - Scopus: 5
High-Performance Real-Time Data Processing: Managing Data Using Debezium, Postgres, Kafka, and Redis
(IEEE, 2023) Çakar, Tuna; Ertuğrul, Seyit; Arslan, Şuayip; Sayar, Alperen; Akçay, Ahmet
This research focuses on monitoring and transferring logs of operations performed on a relational database, specifically PostgreSQL, in real-time using an event-driven approach. The logs generated from database operations are transferred using Apache Kafka, an open-source message queuing system, and Debezium running on Kafka, to Redis, a non-relational (No-SQL) key-value database. Time-consuming query operations and read operations are performed on Redis, which operates on memory (in-memory), instead of on the primary database, PostgreSQL. This approach has significantly improved query execution performance, data processing time, and backend service performance. The study showcases the practical application of an event-driven approach using Debezium, Kafka, Redis, and relational databases for real-time data processing and querying.
Model for Estimating the Probability of a Customer To Have a Transaction
(IEEE, 2022) Sayar Alperen; Çakar Tuna; Ertugrul Seyit; Bozkan Tunahan; Sayar, Alperen; Cakar, Tuna; Ertugrul, Seyit; Bozkan, Tunaban
In this study, it is aimed to estimate the probability of a customer who comes to the institution for the first time to make a transaction in the next 3 months, using data-driven machine learning models, in order to provide financing to the seller company by assigning the receivables arising from the sale of goods and services in a company actively operating in the factoring sector. Accordingly, it was aimed to directly contribute to the transaction volume on a business basis by acting and taking action with more effective, efficient and correct approaches by finding high-potential and low-potential customers. In this context, provided by KKB (Credit Registration Bureau); The data set to he used in machine learning models was created with feature engineering and exploratory data analysis, using the Risk, Mersis, GIB information of the prospective customers and the historical information of the customers, check issuers, customer representatives and branches kept in the database. Since the leads coming to the institution are in two different types of organizations (Individual and Legal), two different forecasting models were applied. Multiple classification models were tried, and the highest F1-Score of 86% for private companies was obtained with the Random Forest model, and the highest F1- Score for commercial companies was obtained with the Random Forest model with 82%. © 2022 IEEE.
Citation - Scopus: 1
Modeling Consumer Creditworthiness Via Psychometric Scale and Machine Learning
(IEEE, 2022) Çakar, Tuna; Ertugrul, Seyit; Sayar, Alperen; Sahin, Türkay; Bozkan, Tunahan
Although the predictive power of economic metrics to detect the creditworthiness of the customers is high, there is a rising interest in the integration of cognitive, psychological, behavioral, alternative, and demographic data into credit risk systems and processing the data through modern methods. The primary motivation for the rising interest is increased customer classification accuracy. In this research, customer creditworthiness was modeled through data consisting of personality, money attitudes, impulsivity, self-esteem, self-control, and material values and processed through artificial intelligence. The obtained findings have been evaluated as a reference point for the following research. © 2022 IEEE.
Multi-Output Vs Single-Output Deep Learning for Plant Disease Detection
(Institute of Electrical and Electronics Engineers Inc., 2025) Taha Kara H.B.; Sayar A.; Gunes P.; Guvencli M.; Ertugrul S.; Cakar T.; Sayar, Alperen; Taha Kara, Hasan Bedri; Guvencli, Mert; Cakar, Tuna; Ertugrul, Seyit; Gunes, Peri
AI-based image processing plays a crucial role in agriculture by enabling early detection of plant diseases, thereby increasing crop productivity and minimizing economic losses. In this study, we present a comparative analysis between a multi-output deep learning model, which simultaneously classifies plant species and assesses their health status, and two separate single-output models trained for each task individually. The publicly available PlantVillage dataset was used for training and evaluation. Performance metrics such as classification accuracy, F1 score, training time, and confusion matrices were used to assess each model. Our results indicate that the multi-output architecture achieves remarkably high classification performance (Plant: 99.98%, Health: 99.78%) while significantly reducing training time by nearly 50% compared to the combined cost of training two individual models. This demonstrates that a unified model not only provides computational efficiency but also maintains predictive strength, making it a practical alternative for real-time agricultural decision support systems. The findings suggest that integrated modeling can contribute to the development of scalable, resource-efficient solutions in precision agriculture. © 2025 IEEE.
Performing Disc Personal Inventory Analysis in Job Postings Using Artificial Intelligence Methods
(Data science and applications, 2023) Sayar, Alperen; Yıldız, Ahmet; Çakar, Tuna; Şengüloğlu, Dilara; Ertuğrul, Seyit
One of the application fields of DISC selfevaluation analysis was introduced to predict people's performance and orientation in their working life. Each letter in the word DISC represents an essential personal characteristic, dividing the profiles of people in business life into four essential parts. In the current study, DISC analysis is conducted on job postings to match the person with the job posting. The current study was based on the analysis of 3 different datasets with job postings in English, Turkish and Romanian prepared by using web scraping methods and then labeled in accordance with DISC criteria. Several different machine learning algorithms have been performed on the DISC analysis outputs, and they reached the best results with accuracy values of around over 96% on the English dataset, around over 95% on the Turkish dataset, and around over 96% on the Romanian dataset, for both D, I, S, C models.
Predicting Credit Repayment Capacity With Machine Learning Models
(Ieee, 2024) Filiz, Gozde; Bodur, Tolga; Yaslidag, Nihal; Sayar, Alperen; Çakar, Tuna
This study examines the transformation in the financial services sector, particularly in banking, driven by the rapid development of technology and the widespread use of big data, and its impact on credit prediction processes. The developed credit prediction model aims to more accurately predict customers' credit repayment capacities. In pursuit of this goal, demographic and financial data along with credit histories of customers have been utilized to employ data preprocessing techniques and test various classification algorithms. Findings indicate that models developed with XGBoost and CATBoost algorithms exhibit the highest performance, while the effective use of feature engineering techniques is revealed to enhance the model's accuracy and reliability. The research highlights the potential for financial institutions to gain a competitive advantage in risk management and customer relationship management by leveraging machine learning models.
Prediction of Loan Decisions With Optical Neuroimaging (fnirs) and Machine Learning
(IEEE, 2023) Girişken, Yener; Son Turan, Semen; Çakar, Tuna; Ertuğrul, Seyit; Sayar, Alperen; Son, Semen; Giriken, Yener
The successful applications of neuroscientific methods and artificial learning approaches have increased in applied fields such as economics, marketing, and finance in the last decade. In this study, a prediction model was developed using the output of optical neuroimaging (fNIRS) measurements from the prefrontal brain regions while 40 participants made decisions for 35 credit offers. The aim was to predict participants' responses to credit offers using artificial learning methods based on four metrics obtained over time from the optical neuroimaging system. The findings of the study indicate that the first 6 seconds (prior to the response entry) are particularly critical. While the performance rate in the developed prediction models is found to be higher, especially in tree-based algorithms, this paper includes a performance comparison of 5 models specifically.
A Predictive Model for Bounced Check Risk Using Machine Learning
(Institute of Electrical and Electronics Engineers Inc., 2025) Kaya K.; Sayar A.; Memis E.C.; Ozlem S.; Ertugrul S.; Cakar T.; Sayar, Alperen; Cakar, Tuna; Ertugrul, Seyit; Ozlem, Sirin; Memis, Emir Cetin; Kaya, Kerem
Bounced checks result in direct monetary losses. Traditional rule-based systems cannot adapt to new patterns and lack flexibility. In this study, we used a large and imbalanced check dataset with customer profiles, credit limits, and historical check outcomes. We applied feature engineering emphasizing time-based transaction patterns, extensive clustering, anomaly detection, and inflation adjustment. We trained six models each for two datasets, which are undersampled to handle class imbalance: Logistic Regression, Random Forest, XGBoost, LightGBM, Extra Trees, and CatBoost. The best performing model, CatBoost, achieved macro F1 scores of 88.5 percent on individual checks dataset with a gross sunk rate of 4.92 percent, and 91.7 percent on corporate checks dataset with a gross sunk rate of 4.28 percent. These results show the model can identify checks most likely to bounce before granting and maintain a low gross sunk rate overall. This study presents a data-driven machine learning solution that enables financial companies to predict and prevent bounced checks before they occur. © 2025 IEEE.
Citation - Scopus: 3
Segmentation for Factoring Customers: Using Unsupervised Machine Learning Algorithms
(IEEE, 2023) Yalçuva, Berat; Akçay, Ahmet; Ertuğrul, Seyit; Çakar, Tuna; Sayar, Alperen; Ayyıldız, Nur Seher
Nowadays the fact that technology facilitates data collection is an important opportunity, as well as making the management of all this data difficult and makes no sense unless it is well processed. This stored data is extremely important, and companies use data provided by their customers. Catching the needs of the customer profiles of the changing world is now a necessity and takes the first place for companies. With the increase in the amount of stored data over time, it has become difficult to establish a relationship between the data and to separate them from each other. At this point, machine learning methods have become more involved in our lives. In this study, what segmentation is and its change over the years are mentioned. It has been mentioned which machine learning techniques will be useful in data selection. Then, possible machine learning methods are shown in real life segmentation problem by using the domestic factoring company’s customer check data. Since this study aims to group unlabeled data, unsupervised learning techniques are emphasized. Among these methods, Hierarchical Clustering, DBSCAN, Gaussian Mixture Modeling methods, Fuzzy c- Means were used as well as the most popular K-Means algorithm. When the clustering results were examined, the optimal number of clusters was calculated very high with GMM, DBSCAN could not assign clusters, and Hierarchical clustering could not produce expected results. It was observed that the best results were obtained with the K-Means and Fuzzy c - Means algorithms.
Citation - Scopus: 2
Ssqem: Semi-Supervised Quantum Error Mitigation
(IEEE, 2022) Sayar, Alperen; Arslan Suayb S.; Çakar Tuna; Arslan, Suayb S.; Cakar, Tuna
One of the fundamental obstacles for quantum computation (especially in noisy intermediate-scale quantum (NISQ) era) to be a near-term reality is the manufacturing gate/measurement technologies that make the system state quite fragile due to decoherence. As the world we live in is quite far away from the ideal, complex particle-level material imperfections due to interactions with the environment are an inevitable part of the computation process. Hence keeping the accurate state of the particles involved in the computation becomes almost impossible. In this study, we posit that any physical quantum computer sys-tem manifests more multiple error source processes as the number of qubits as well as depth of the circuit increase. Accordingly, we propose a semi-supervised quantum error mitigation technique consisting of two separate stages each based on an unsupervised and a supervised machine learning model, respectively. The proposed scheme initially learns the error types/processes and then compensates the error due to data processing and the projective measurement all in the computational basis. © 2022 IEEE.