KaydedilenlerDerginizi ekleyin

Dergiler / Afyon Kocatepe Üniversitesi İktisadi ve İdari Bilimler Fakültesi Dergisi / 2026 / Cilt: 28 Sayı: 1

Klasik Makine Öğrenimi ile Finansal Haber Duygu Sınıflandırması: Tekrarlanabilir ve Maliyet Etkin Bir Ensemble Yaklaşımı

Financial news sentiment classification with classical machine learning: A reproducible and cost-effective Ensemble approach

Kamil Abdullah Eşidir, Emre Ünal, Yunus Emre Gür

Sayfa
159–178
Yayın Tarihi
2026-03-23
🏛️Bu makalenin bibliyografik verileri ve tam metin bağlantısı DergiPark üzerinden sağlanmıştır.
DergiPark Kaydı ↗

Özet

Bu çalışma, finansal haber metinlerinin olumlu, olumsuz ve nötr sınıflara ayrılmasını amaçlamaktadır. Çalışmanın temel hedefi, GPU desteği gerektirmeyen koşullarda yeniden üretilebilir bir sınıflandırma modeli geliştirmek ve klasik makine öğrenmesi algoritmalarının performansını karşılaştırmaktır. Analizlerde Kaggle üzerinde yayımlanan “Finance News Sentiments” isimli veri kümesi kullanılmıştır. Veri kümesi 32.417 adet haber başlığından oluşmakta ve üç duygu sınıfında da dengeli bir dağılım göstermektedir. Veri ön işleme sürecinde metin temizleme, kök bulma ve durdurma sözcüklerin çıkarılması adımları uygulanmış; TF-IDF tabanlı vektörleştirme yöntemi ile metinler sayısal forma dönüştürülmüştür. Lojistik Regresyon, Multinomial Naive Bayes, XGBoost ve Ensemble modelleri beş katlı çapraz doğrulama ile test edilmiştir. Performans değerlendirmesinde doğruluk, F1-Skoru ve ROC AUC ölçütleri kullanılmıştır. Bulgular, doğruluk oranlarının %59–65 arasında değiştiğini göstermiştir. Ensemble yaklaşımı %65,3 doğruluk, %65,2 Makro F1-Skoru ve 0,83 ROC AUC ile en başarılı sonucu vermiştir. Ayrıca SHAP analizi, özniteliklerin model kararlarındaki etkisini ortaya koymuş ve açıklanabilirlik boyutunu güçlendirmiştir. Sonuçlar, klasik makine öğrenmesi yöntemlerinin düşük donanım koşullarında dahi istikrarlı performans sergileyebildiğini göstermektedir. Çalışma, TF-IDF tabanlı temsillerin kısa haber metinlerinde yüksek etkililik sunduğunu ve maliyet–doğruluk dengesi bağlamında literatüre özgün katkı sağladığını ortaya koymaktadır.

Abstract

This study aims to classify financial news texts into positive, negative, and neutral categories. The main objective is to develop a reproducible classification model that does not require GPU support and to compare the performance of classical machine learning algorithms. The analysis uses the Finance News Sentiments dataset published on Kaggle, which contains 32,417 news headlines with a balanced distribution across three sentiment classes. In the preprocessing stage, text cleaning, stemming, and stop-word removal steps were applied; then, the texts were transformed into numerical form using TF-IDF vectorization. Logistic Regression, Multinomial Naive Bayes, XGBoost, and Ensemble models were tested through five-fold cross-validation. Model performance was evaluated with Accuracy, F1-Score, and ROC AUC metrics. The findings show that accuracy rates ranged between 59% and 65%. The ensemble approach achieved the best performance with 65.3% accuracy, a 65.2% Macro F1-Score, and 0.83 ROC AUC. In addition, SHAP analysis revealed the contribution of features to model decisions and strengthened interpretability. Overall, the results demonstrate that classical machine learning methods can provide stable performance even under low-resource conditions. The study shows that TF-IDF-based representations are highly effective in short news texts and provides an original contribution to the literature in terms of the cost–accuracy trade-off (the balance between accuracy and computational cost).

Anahtar kelimeler
🔖

Bu Makaleye Atıf Yap

Eşidir, K. A., Ünal, E., Gür, Y. E. (2026). Klasik Makine Öğrenimi ile Finansal Haber Duygu Sınıflandırması: Tekrarlanabilir ve Maliyet Etkin Bir Ensemble Yaklaşımı. Afyon Kocatepe Üniversitesi İktisadi ve İdari Bilimler Fakültesi Dergisi, Cilt: 28 Sayı: 1, 159–178. https://doi.org/10.33707/akuiibfd.1791917

📄 Tam Metin PDF Okuyucu

Kaynakça (50)

  1. Abar, H. (2020). XGBoost ve Mars yöntemleriyle altın fiyatlarının kestirimi. EKEV Akademi Dergisi, 83, 427-446.
  2. Adak, A., Pradhan, B., ve Shukla, N. (2022). Sentiment Analysis of Customer Reviews of Food Delivery Services Using Deep Learning and Explainable Artificial Intelligence: Systematic Review. Foods, 11(10), 1500. https://doi.org/10.3390/foods11101500
  3. Adhikari, S., Thapa, S., Naseem, U., Lu, H. Y., Bharathy, G., ve Prasad, M. (2023). Explainable hybrid word representations for sentiment analysis of financial news. Neural Networks, 164, 115-123. https://doi.org/10.1016/j.neunet.2023.04.0114
  4. Agarwal, A. (2020, September). Sentiment analysis of financial news. In 2020 12th International Conference on Computational Intelligence and Communication Networks (CICN), pp. 312-315, Bhimtal, India. https://doi.org/10.1109/CICN49253.2020.9242579
  5. Ahmad, H. O. ve Umar, S. U. (2023). Sentiment analysis of financial textual data using machine learning and deep learning models. Informatica, 47(5). https://doi.org/10.31449/inf.v47i5.4673
  6. Almalis, I., Kouloumpris, E., ve Vlahavas, I. (2022). Sector-level sentiment analysis with deep learning. Knowledge-Based Systems, 258, 109954. https://doi.org/10.1016/j.knosys.2022.109954
  7. Aslan, K. (2025). Yapay Zekâ, Makine Öğrenmesi ve Veri Bilimi Kursu, Sınıfta Yapılan Örnekler ve Özet Notlar, C ve Sistem Programcıları Derneği, İstanbul.
  8. Benedetti, A. (2024), “Finance News Sentiments Data Set”, Kaggle, Erişim Adresi: https://www.kaggle.com/datasets/antobenedetti/finance-news-sentiments, Erişim Tarihi: 15.06.2025
  9. Bi, J. (2022). Stock market prediction based on financial news text mining and investor sentiment recognition. Mathematical Problems in Engineering, 2022(1), 2427389. https://doi.org/10.1155/2022/2427389
  10. Chavakula, S. J., Albert, C. A. J., Ebenezer, E., Bhagat, M. H., ve Mahamuni, C. V. (2025, March). Explainable AI (XAI) Using SHAP and LIME for Financial Fraud Detection and Credit Scoring. In 2025 International Conference on Advanced Computing Technologies (ICoACT), 1-8, IEEE, Sivalasi, India. https://doi.org/10.1109/ICoACT63339.2025.11005238
  11. Chen, P., Boukouvalas, Z., ve Corizzo, R. (2024). A deep fusion model for stock market prediction with news headlines and time series data. Neural Computing and Applications, 36(34), 21229-21271. https://doi.org/10.1007/s00521-024-10303-1
  12. Chen, T., ve Guestrin, C. (2016). XGBoost: A scalable tree boosting system. Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 785–794. https://doi.org/10.1145/2939672.2939785
  13. Cheng, Q., Ke, Y., ve AbdelMouty, A. M. (2020). Negative emotion diffusion and intervention countermeasures of social networks based on deep learning. Journal of Intelligent ve Fuzzy Systems, 39(4), 4935-4945. https://doi.org/10.3233/jifs-179979
  14. Çiçek, A., ve Arslan, Y., 2020. Müşteri Kayıp Analizi İçin Sınıflandırma Algoritmalarının Karşılaştırılması. İleri Mühendislik Çalışmaları Ve Teknolojileri Dergisi, 1(1), 13-19.
  15. Danyal, M. M., Haseeb, M. T., Khan, B., Ullah, S., ve Khan, S. S. (2024). Opinion mining on movie reviews based on deep learning models. Journal on Artificial Intelligence, 6(1), 23-42. https://doi.org/10.32604/jai.2023.045617
  16. Didolkar, A., ve Lokulwar, P. (2025, January). Sentiment Detection in Financial News: A Deep learning Approach to Extreme and Moderate Classification. In 2025 International Conference on Intelligent and Innovative Technologies in Computing, Electrical and Electronics (IITCEE), 1-8, Bangalore, India. https://doi.org/10.1109/IITCEE64140.2025.10915503
  17. Diwali, A., Saeedi, K., Dashtipour, K., Gogate, M., Cambria, E., ve Hussain, A. (2023). Sentiment analysis meets explainable artificial intelligence: A survey on explainable sentiment analysis. IEEE Transactions on Affective Computing. 15(3), 837-846. https://doi.org/10.1109/TAFFC.2023.3296373
  18. Dong, S., ve Liu, C. (2021). Sentiment classification for financial texts based on deep learning. Computational Intelligence and Neuroscience, 2021, 9524705. https://doi.org/10.1155/2021/9524705
  19. Fatouros, G., Soldatos, J., Kouroumali, K., Makridis, G., ve Kyriazis, D. (2023). Transforming sentiment analysis in the financial domain with ChatGPT. Machine Learning with Applications, 14, 100508. https://doi.org/10.1016/j.mlwa.2023.100508
  20. Fazlija, B., ve Harder, P. (2022). Using financial news sentiment for stock price direction prediction. Mathematics, 10(13), 2156.
  21. Gite, S., Khatavkar, H., Kotecha, K., Srivastava, S., Maheshwari, P., ve Pandey, N. (2021). Explainable stock prices prediction from financial news articles using sentiment analysis. PeerJ Computer Science, 7, e340. https://doi.org/10.7717/peerj-cs.340
  22. Hossin, M., ve Sulaiman, M. N. (2015). A review on evaluation metrics for data classification evaluations. International journal of data mining ve knowledge management process, 5(2), 1-11. https://doi.org/10.5121/ijdkp.2015.5201
  23. Huang, B. ve Wei, J. (2021). Huang, B., ve Wei, J. (2021). Research on Deep Learning‐Based Financial Risk Prediction. Scientific programming, 2021(1), 6913427. https://doi.org/10.1155/2021/6913427
  24. Jun Gu, W., hao Zhong, Y., zun Li, S., song Wei, C., ting Dong, L., yue Wang, Z., ve Yan, C. (2024, August). Predicting stock prices with finbert-lstm: Integrating news sentiment analysis. In Proceedings of the 2024 8th International Conference on Cloud and Big Data Computing, pp. 67-72. https://doi.org/10.1145/3694860.3694870
  25. Kayakuş, M. ve Yiğit Açıkgöz, F. (2023). Twitter'da Makine Öğrenmesi Yöntemleriyle Sahte Haber Tespiti. Abant Sosyal Bilimler Dergisi, 23(2), 1017-1027. https://doi.org/10.11616/asbi.1266179
  26. Kilimci, Z. H. and Duvar, R. (2020). An efficient word embedding and deep learning based model to forecast the direction of stock exchange market using twitter and financial news sites: a case of istanbul stock exchange (bist 100). IEEE Access, 8, 188186-188198. https://doi.org/10.1109/access.2020.3029860
  27. Liu, N., Sekiguchi, K., Nakata, Y., Sugie, T., Yoshino, T., ve Ohsawa, Y. (2025). Hierarchical Structure Estimation of Financial Networks and Sentiment Analysis of Financial News Headlines. in IEEE Access, vol. 13, pp. 129476- 129492, https://doi.org/10.1109/ACCESS.2025.3588385
  28. Madhuri, A., Divya, K. N., Varshini, K., Harsha, K. B., Pujitha, K., ve Praveen, K. N. (2025, March). News Categorization Using Least Squares Twin Support Vector Machine. In 2025 International Conference on Visual Analytics and Data Visualization (ICVADV, 1547-1553, Tirunelveli, India. http://doi.org/10.1109/ICVADV63329.2025.10960905
  29. Mahendran, M. B., Gokul, A. K., Lakshmi, P., ve Pavithra, S. (2025, February). Comparative Advances in Financial Sentiment Analysis: A Review of BERT, FinBert, and Large Language Models. In 2025 3rd International Conference on Intelligent Data Communication Technologies and Internet of Things (IDCIoT), pp. 39-45, IEEE, Bengaluru, India. https://doi.org/10.1109/IDCIOT64235.2025.10914764
  30. Mamun, A. A., Hossain, M. S., Rishad, S. S. I., Rahman, M. M., Shakil, F., Choudhury, M. Z. M. E., Das, A. C., Das, R., ve Sultana, S. (2024). Machine learning for stock market security measurement: A comparative analysis of supervised, unsupervised, and deep learning models. The American Journal of Engineering and Technology, 6(11), 63-76. https://doi.org/10.37547/tajet/Volume06Issue11-08
  31. Mehta, P., Pandya, S., ve Kotecha, K. (2021). Harvesting social media sentiment analysis to enhance stock market prediction using deep learning. PeerJ Computer Science, 7, e476. https://doi.org/10.7717/peerj-cs.476
  32. Memiş, E., Akarkamçı, H., Yeniad, M., Rahebi, J., ve Lopez-Guede, J. M. (2024). Comparative Study for Sentiment Analysis of Financial Tweets with Deep Learning Methods. Applied Sciences, 14(2), 588. https://doi.org/10.3390/app14020588
  33. More, S. V. (2024). Stock Market Prediction Using Financial News Sentiments and Technical Indicator Data with Machine Learning Models and LIME for Explainable Insights (Doctoral dissertation, Dublin, National College of Ireland).
  34. Muthukumar, P., ve Zhong, J. (2021). A stochastic time series model for predicting financial trends using nlp. arXiv preprint arXiv:2102.01290. https://doi.org/10.48550/arXiv.2102.01290
  35. Pandey, A., Sharma, E., Tiwari, R., Bisht, A., ve Prathaban, B. P. (2025, April). Precision Driven Sentiment and Topic Classification of News Articles using IndicBert. In 2025 International Conference on Recent Advances in Electrical, Electronics, Ubiquitous Communication, and Computational Intelligence (RAEEUCCI), pp. 1-6, IEEE, Chennai, India, https://doi.org/10.1109/RAEEUCCI63961.2025.11048217
  36. Popoola, G., Abdullah, K. K., Fuhnwi, G. S., ve Agbaje, J. (2024, February). Sentiment analysis of financial news data using tf-idf and machine learning algorithms. In 2024 IEEE 3rd International Conference on AI in Cybersecurity (ICAIC), pp. 1-6, IEEE, Houston, TX, USA, https://doi.org/10.1109/ICAIC60265.2024.10433843
  37. Saxena, A., Santhanavijayan, A., Shakya, H. K., Kumar, G., Balusamy, B., ve Benedetto, F. (2024). Nested Sentiment Analysis for ESG Impact: Leveraging FinBERT to Predict Market Dynamics Based on Eco-Friendly and Non-Eco-Friendly Product Perceptions with Explainable AI. Mathematics, 12(21), 3332. https://doi.org/10.3390/math12213332
  38. Seshakagari, H. R. B., Umashankar, A., Harikala, T., Jayasree, L., ve Severance, J. (2025). Dynamic financial sentiment analysis and market forecasting through large language models. International Journal of Human Computations ve Intelligence, 4(1), 397-410. https://doi.org/10.5281/zenodo.15111609
  39. Sinha, A., Kedas, S., Kumar, R., ve Malo, P. (2022). SEntFiN 1.0: Entity‐aware sentiment analysis for financial news. Journal of the Association for Information Science and Technology, 73(9), 1314-1335. https://doi.org/10.1002/asi.24634
  40. Song, Y., Liu, X., ve Ze, Z. (2024). A comprehensive review of text classification algorithms. Journal of Electronics and Information Science, 9(2). https://doi.org/10.23977/jeis.2024.090205
  41. Štrimaitis, R., Stefanovič, P., Ramanauskaitė, S., ve Slotkienė, A. (2021). Financial Context News Sentiment Analysis for the Lithuanian Language. Applied Sciences, 11(10), 4443. https://doi.org/10.3390/app11104443
  42. Sun, W., Wang, Y., Zhang, L., Chen, X. H., ve Hoang, Y. H. (2025). Enhancing economic cycle forecasting based on interpretable machine learning and news narrative sentiment. Technological Forecasting and Social Change, 215, 124094. https://doi.org/10.1016/j.techfore.2025.124094
  43. Taheripour, E., Sadjadi, S. J., ve Amiri, B. (2025). A Novel Approach to Portfolio Construction: An Application of FinBERT Sentiment Analysis and Credibilistic CVaR Criterion. IEEE Access. vol. 13, pp. 76775-76795, https://doi.org/10.1109/ACCESS.2025.3564615
  44. Trisna, I. N. P., Dewi, N. W. E. R., ve Pasirulloh, M. A. (2025). Oversampling vs. undersampling in TF-IDF variations for imbalanced Indonesian short texts classification. TELKOMNIKA (Telecommunication Computing Electronics and Control), 23(2), 382-392. http://doi.org/10.12928/telkomnika.v23i2.26510
  45. Vaca, C., Astorgano, M., Lopez-Rivero, A. J., Tejerina, F., ve Sahelices, B. (2024). Interpretability of deep learning models in analysis of Spanish financial text. Neural Computing and Applications, 36(13), 7509-7527. https://doi.org/10.1007/s00521-024-09474-8
  46. Wajeed, M. A., Quadry, K. M., Mallik, M. A., ve Khanna, K. R. (2025). Explainable Machine Learning For Text Classifıcation: A Novel Approach To Transparency And Interpretability. Journal of Theoretical and Applied Information Technology, 103(3).
  47. Wang, L., Wang, X., Chen, A., Jin, X., ve Che, H. (2020). Prediction of type 2 diabetes risk and its effect evaluation based on the XGBoost model. Healthcare, 8(3), 247. https://doi.org/10.3390/healthcare8030247
  48. Xiao, P. (2024). Stock market prediction based on financial news, text data mining, and investor sentiment analysis. International Journal of Information System Modeling and Design, 15(1), 1-13. https://doi.org/10.4018/ijismd.361593
  49. Yang, J., Cai, Q., Gong, J., ve Han, M. C. (2025). The financial institution text data mining and value analysis model based on big data and natural language processing. Journal of Organizational and End User Computing, 37(1), 1-40. https://doi.org/10.4018/joeuc.374213
  50. Zhang, Y. (2025). Interpretable Machine Learning for Macro Alpha: A News Sentiment Case Study. arXiv preprint arXiv:2505.16136. https://doi.org/10.48550/arXiv.2505.16136