Sürdürülebilir Kalkınma Amaçları
Yayınlar
- BildiriYalnız künye2025Leveraging RFM Features for Churn Prediction in Imbalanced E-Commerce Data: A Hepsiburada Case Study
Demirtas, Sude Naz; Evren, Erhan; Ozdemir, Murat; Hacibeyoglu, Elif Pinar; Vural, Ulas; Adali, E.
The share of the e-commerce sector in the global economy and the level of market competition are rapidly increasing. Retaining existing customers generally incurs lower costs than acquiring new ones, making churn prediction a strategic priority for e-commerce platforms. This study introduces a framework that leverages RFM (Recency, Frequency, Monetary) scores calculated across 32 product categories to construct symbolic behavioral feature vectors for churn prediction. Unlike traditional RFM segmentation, these categorical scores are directly used as inputs for supervised machine learning models, enhancing interpretability and enabling fine-grained behavioral analysis. A binary churn label was generated based on purchase activity over the final two months of a 10-month period in 2024, using a real-world e-commerce dataset from HepsiBurada.com, revealing significant class imbalance (82% active, 18% churned). To address this, seven resampling strategies - including SMOTE variants and hybrid techniques - were applied in combination with five classifiers: Decision Tree, Random Forest, Support Vector Machine, XGBoost, and Artificial Neural Network. A total of 35 model-resampling combinations were evaluated using accuracy, F1-score, recall, and confusion matrices. Random Forest combined with Random Oversampling achieved the highest performance, with 95% accuracy and 94% F1-score. The representation of RFM-based features proved particularly robust under imbalanced conditions of a real-world e-commerce dataset from HepsiBurada.com. The proposed framework demonstrates the effectiveness of integrating RFM-based segmentation with predictive modeling for real-world e-commerce scenarios and provides actionable insights for businesses aiming to proactively identify at-risk customers and reduce churn.
- BildiriYalnız künye2025Risk Estimation for Financial Contracts: Data Imbalance Elimination and Impact of Different Data Rates
Caliskan, Elif Ezel; Yildizak, Burak; Ozcan, Dogan; Salbacak, Gurbet Alpay; Vural, Ulas; Aydin, Tarkan; Adali, E.
Predicting the risk score of financial contracts plays a critical role in financial risk management by enabling companies to make more informed decisions. When predicting risks based on contracts, several challenges arise. In the financial sector, the imbalance of data is particularly common. This study comprehensively investigates the impact of the class imbalance problem on calculating the risk scores of financial contracts. To improve the performance of machine learning methods on imbalanced data, techniques such as Synthetic Minority Oversampling Technique (SMOTE), Adaptive Synthetic Sampling Approach for Imbalanced Learning (ADASYN), Tabular Variational Autoencoder (TVAE), and Conditional Tabular Generative Adversarial Network (CTGAN) have been used for data balancing. Logistic Regression (LR) and Gradient Boosting (GB) models have been employed to evaluate the effect of these techniques on model performance. The main objective of the study is to analyze the performance of machine learning models based on different data imbalance ratios and assess their effectiveness in risk prediction.
- BildiriYalnız künye2024Machine Learning Approaches to Predict Thyroid Cancer Recurrence: A Comparative Study
Ozturk, Candide; Sagir, Ozgur; Vural, Ulas; Adali, E.
Thyroid cancer recurrences have significant impacts on patients' health. The current clinical methods have limited prediction accuracy because of inadequate incorporation of individual patient variability. Therefore, there is a need for advanced techniques that can help medical experts to make accurate and reliable predictions. Recent efforts on the topic are focused on machine learning algorithms. This study aims to assess the performance of these algorithms for predicting thyroid cancer recurrence. Four prediction models are trained using Gradient Boosting, Random Forest, XGBoost, and AdaBoost algorithms. The performances of the models are evaluated with metrics such as accuracy, F1 score, F-05 score, and area under the ROC curve (AUC). Experiments with the K-Fold (5-fold) method have revealed how effective these models can be in the prediction of thyroid cancer recurrence. The performance of the machine learning models was evalu-ated using accuracy, Fl-score, F05-score, and AUC score metrics. The XGBoost model outperformed others, achieving an accuracy of 95,3%, Fl-score of 91,8%, F05-score of 91,9%, and AUC score of 98,2%. These results show that machine learning-based solutions may contribute to incorporating more accurate and individualized treatment decisions in clinical practice.
AtıfScopus6OpenAlex6
İncele