Browsing by Author "Ertugrul, Seyit"

Now showing 1 - 9 of 9

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.
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.
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.
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.
Yapay Öğrenme Tabanlı Mikrofaktoring Skorlama Modeli ve Kredi Risk Yönetim Sistemi Geliştirilmesi
(Institute of Electrical and Electronics Engineers Inc., 2025) Sayar, Alperen; Ates, Yigit; Ertugrul, Seyit; Turan, Elif Naz; Drias, Yassine; Cakar, Tuna
Credit scoring systems are critical tools used by factoring institutions to assess the credit risks of SME businesses seeking microloans. This study presents a comprehensive predictive modeling framework that achieves 82.67% ROC-AUC with 65.34% Gini score on test data, demonstrating robust discriminative capability despite significant class imbalance. Our ensemble approach outperforms individual boosting models by leveraging their complementary strengths in payment behavior analysis and fraud detection. The raw data was cleaned, transformed, and optimized using the Polars library, with specialized features for detecting fraud patterns and time-based risk indicators. When implementing a score threshold of 950, our model significantly improves the detection of non-performing loans (NPL) compared to traditional rule-based approaches by reducing the net deficit from 6.59% to 2.62%. When applied to previously rejected applications, the model projects a potential 762.57% increase in transaction count and 747.05% growth in transaction volume. © 2025 Elsevier B.V., All rights reserved.