Türkiye’de Cari İşlemler Dengesinin Aylık Tahmininde Makine Öğrenmesi ve Derin Öğrenme Yöntemleri: Karşılaştırmalı Bir Analiz

Yıl 2026, Cilt: 26 Sayı: 1, 1 - 21, 27.03.2026

Hakan Öndes

https://doi.org/10.11616/asbi.1742565

https://izlik.org/JA66WU57NA

Öz

Bu çalışma, Ocak 2013 – Nisan 2025 dönemine ait aylık veriler kullanılarak, Türkiye’nin cari işlemler dengesini makine öğrenmesi (SVR, XGBoost) ve derin öğrenme (RNN, LSTM, GRU) modelleriyle tahmin etmeyi amaçlamaktadır. İşsizlik, döviz kuru, faiz oranı ve dış ticaret dengesi gibi on bir makroekonomik değişken bağımsız değişken olarak kullanılmıştır. GRU ve LSTM modelleri, doğruluk açısından diğer modellere üstünlük sağlamış; GRU en düşük MAE ve en yüksek R² değerlerini üretmiştir. Tüm modeller çapraz doğrulama, normalizasyon ve hiperparametre ayarlarıyla test edilmiştir. Sonuçlar, bellek tabanlı sinir ağlarının makroekonomik zaman serilerinin dinamik ve doğrusal olmayan yapısını etkili biçimde yakalayabildiğini göstermektedir. Bu çalışma, Türkiye’de cari denge tahmini için derin öğrenmenin uygulandığı ilk çalışmalardan biridir ve veri odaklı ekonomi politikalarının tasarımı açısından önemli çıkarımlar sunmaktadır.

Anahtar Kelimeler

Makine ve Derin Öğrenme , Cari İşlemler Dengesi , Türkiye Ekonomisi

Machine Learning and Deep Learning Methods in Monthly Current Account Balance Predicting in Turkey: A Comparative Analysis

https://izlik.org/JA66WU57NA

Öz

This study aims to predict Turkey’s current account balance using machine learning (SVR, XGBoost) and deep learning models (RNN, LSTM, GRU) based on monthly data from January 2013 to April 2025. Eleven macroeconomic variables including unemployment, exchange rates, interest rates, and foreign trade balance were used as predictors. The GRU and LSTM models outperformed others in terms of accuracy, with GRU yielding the lowest MAE and highest R². All models were evaluated using cross-validation, normalization, and hyperparameter tuning. Results show that memory-based neural networks can effectively capture the dynamic and nonlinear structure of macroeconomic time series. This study is one of the first to apply deep learning to Turkey’s current account forecasting and provides valuable insights for data-driven economic policy design.

Anahtar Kelimeler

Machine and Deep Learning , Current Account Balance , Turkish Economy

Kaynakça

• Akçay, B. (2012). Current Account Deficit Sustainability in Turkey: A Comparison with Greece in Debt Crisis, Ekonomik Yaklasim Association, 23(84), p.1-38. https://www.ekonomikyaklasim.org/fulltext/94-1390663332.pdf
• Akkaya, M. (2022). Analysis of the Factors Affecting the Current Account Balance: Turkey Case. Journal of Management and Economics, 29(4), p.707-722. https://dergipark.org.tr/en/download/article-file/2673327
• Altunöz, U. (2014). Fundamental Reasons of Current Deficit and Sustainabilty: The Case of Turkey. Istanbul Gelisim University Journal of Social Sciences, 2, p.118-122. https://dergipark.org.tr/tr/download/article-file/89255
• Aristovnik, A. (2007). Short – and Medium- Term Determinants of Current Account Balances in Middle East and North Africa Countries, The William Davidson Institute Working Paper, The University of Michigan, No: 862. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=988171
• Barışık, S. and Çetintaş, H. (2006). Current Deficits Sustainability in Turkey: A Structural Break Model, 1987-2003. Süleyman Demirel University Journal of Faculty of Economics and Administrative Sciences, 11(1), p.1-16. https://dergipark.org.tr/tr/download/article-file/194880
• Batdelger, T., and Kandil, M. (2011). Determinants of the current account balance in the United States. Applied Economics, 44(5), p.653–669. https://doi.org/10.1080/00036846.2010.518950
• Bayar, Y., Kılıç, C. and Arıca, F. (2014), Determinants of Current Account Deficits in Turkey, Cumhuriyet University Journal of Economics and Administrative Sciences, 15(1), p.451-471. http://esjournal.cumhuriyet.edu.tr/en/download/article-file/48541
• Berke, B. (2009). Sustainability of The Current Deficit in Turkey: Fractional Cointegration Analysis, Akdeniz University Faculty of Economics and Administrative Sciences, 9 (18), p.117-145. https://dergipark.org.tr/en/download/article-file/372668
• Bozgeyik, Y. and Aydın, K. (2019). Determinants of Current Account Deficit in Turkey: An Empirical Analysis for 1992-2017. Finance Journal , 176: p. 1-26. https://ms.hmb.gov.tr/uploads/2019/09/176-01.pdf
• Brissimis, S.N., Hondroyiannis, G. and Papazoglou, C. (2013). The determinants of current account imbalances in the euro area: a panel estimation approach. Econ Change Restruct 46, p.299–319. https://doi.org/10.1007/s10644-012-9129-0
• Chen,T. ve Guestrin,C.(2016). XGBoost: A scalable tree boosting system. Paper presented at the Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. p.785-794. Retrieved from https://doi.org/10.114 5/2939672.2939785
• Colak, Z. (2025). Stock Price Prediction with Deep Learning Models: Comparative Analysis of Lstm, Gru, Rnn, Mlp Models. Journal of Management Sciences, 23(56), p.1250-1286. https://dergipark.org.tr/tr/download/article-file/4416982
• Das, D.K. (2016). Determinants of current account imbalance in the global economy: a dynamic panel analysis. Economic Structures, 5(8), p.1-24. https://doi.org/10.1186/s40008-016-0039-6
• Erdoğan,S. and Bozkurt,H. (2009). The Determinants of Current Account Deficit in Turkey: An Analysis with MGARCH Models, Journal of Finance Letters, 23(84), p.135-172. https://dergipark.org.tr/en/download/article-file/150781
• Fischer, T. and Krauss, C. (2018). Deep learning with long short-term memory networks for financial market predictions, European Journal of Operational Research, 270(2), p.654-669. https://doi.org/10.1016/j.ejor.2017.11.054
• Gençoğlu, P. and Ünlü, F. (2019). Sustainability of Current Account Deficit in Turkey: An Econometrics Analysis. Hacettepe University Journal of Economics and Administrative Sciences, 37(4): p.627-650. https://dergipark.org.tr/en/download/article-file/911182
• Göçer, İ.(2013). Reasons, Financing Quality and Sustainability of Current Account Deficit in Turkey: An Econometric Analysis, Eskişehir Osmangazi University Journal of Economics and Administrative Sciences, 8(1), p.213-242. https://dergipark.org.tr/en/download/article-file/65435 Gupta, R., Srivastava, D. and Sahu, M. (2021). Artificial intelligence to deep learning: machine intelligence approach for drug discovery. Mol Divers 25, p.1315–1360. https://doi.org/10.1007/s11030-021-10217-3
• Hiransha, M., Gopalakrishnan, E.A., Menon, V.K. and Soman K.P. (2018). NSE Stock Market Prediction Using Deep-Learning Models, Procedia Computer Science, 132, p.1351-1362, https://doi.org/10.1016/j.procs.2018.05.050
• Hopp, D. (2022). Economic Nowcasting with Long Short-Term Memory Artificial Neural Networks (LSTM). Journal of Official Statistics, 38(3), p.847-873. https://doi.org/10.2478/jos-2022-0037
• Kamoltip, S. (2021). Macroeconomic forecasting with LSTM and mixed frequency time series data, http://dx.doi.org/10.48550/arXiv.2109.13777 , arXiv preprint.
• Karış, Ç. (2020). Relationship Between Current Deficit Definitions and Sustainability Probability: Turkey. Anemon Muş Alparslan University Journal of Social Sciences, 8, p.99–106. http://dx.doi.org/10.18506/anemon.63451 0.
• Kaya, M. (2016). Current Account Deficit Problem and Its Causes In Turkey. Dicle University Journal of Economics and Administrative Sciences, 6(10), p.51-75. https://dergipark.org.tr/en/download/article-file/370720
• Kılavuz, E., and Yücer, E. N (2022). An Analysis on Determinants of Current Account Deficit in Turkey with ARDL Bounds Test Approach. Journal of Doğuş University, 23(2), p.251-267. https://dergipark.org.tr/en/download/article-file/2006852
• Kılınç, H. Ç. and Öztürk, Y. (2022). Time Series Forecasting Using a Gated Recurrent Unit Model Hybridized with a Gray Wolf Optimization. Journal of European science and technology, (35), p.259-267. https://doi.org/10.31590/ejosat.1062777
• Koç, S. and Bakırtaş, İ. (2016). Sustainability of Current Account Deficit In Turkey: Evidence From Cointegration Tests. Dumlupınar University, Journal of Social Sciences, (49), p.252-277. https://dergipark.org.tr/en/download/article-file/264429
• Kwalingana, S., and O. Nkuna. (2009). The Determinants of Current Account Imbalances in Malawi. MPRA Ma, X., Sha J., Wang, D., Yu, Y., Yang, Q. and Niu, X. (2018). Study on a prediction of P2P network loan default based on the machine learning LightGBM and XGboost algorithms according to different high dimensional data cleaning, Electronic Commerce Research and Applications. 31, p.24-39. https://doi.org/10.1016/j.elerap.2018.08.002.
• Magazzino, C. and Haroon, M. (2025). The interrelation among environmental quality, public accounts, and macroeconomic fundamentals: An analysis of OECD. Environmental Dvelopment, 54, 1-16. https://doi.org/10.1016/j.envdev.2025.101175.
• Makridakis, S., Spiliotis, E. and Assimakopoulos, V. (2018) Statistical and Machine Learning forecasting methods: Concerns and ways forward. Plos One, 13(3): p.1-26. https://doi.org/10.1371/journal.pone.0194889
• Masini, R. P., Medeiros, M. C., and Mendes, E. F. (2023). Machine learning advances for time series forecasting. J Econ Surv, 37, p. 76–111. https://doi.org/10.1111/joes.12429
• Mohandes, M. (2002), Support vector machines for short‐term electrical load forecasting. International Journal of Energy Research, 26(4), p.335-345. https://doi.org/10.1002/er.787
• Olah, C. (2015). Understanding LSTM networks. http://colah.github.io/posts/2015-08-Understanding-LSTMs/ Sadiku, L., Fetahi-Vehapi, M., Sadiku, M. and Berisha, N. (2015). The Persistence and Determinants of Current Account Deficit of FYROM: An Empirical Analysis, Procedia Economics and Finance, 33, p.90-102. https://doi.org/10.1016/S2212-5671(15)01696-2.
• Sakarya, Ş. and Yılmaz, Ü. (2019). Prediction of BIST30 Index Using Deep Learning Architecture. European Journal of Educational & Social Sciences. 4(2), p.106-121. https://dergipark.org.tr/en/download/article-file/856334
• Santamaría-Bonfil, G., Frausto-Solís, J. and Vázquez-Rodarte, I. (2015) Volatility forecasting using support vector regression and a hybrid genetic algorithm. Computational Economics, 45(1), p.111-133. http://hdl.handle.net/10.1007/s10614-013-9411-x
• Sarıtaş, H., Genç, A. and Avcı, T. (2018). Relationship Between Energy Import, Current Account Deficit and Growth in Turkey: Var and Granger Causality Analysis. The International Journal of Economic and Social Research, 14 (2), p.181-200. https://dergipark.org.tr/en/download/article-file/668600
• Shevchuk V. O., Synchak V., Zaverbnyj A. S. and Baranetska O. V. (2020). Determinants of the Current Account Balance and Output in Ukraine, Financial and Credit Activity Problems of Theory and Practice, 3(30), p.186-195. https://doi.org/10.18371/fcaptp.v3i30.179532
• Smola, A. J. and Schölkopf, B. (2004), A tutorial on support vector regression. Statistics and computing, 14(3), p.199-222. https://doi.org/10.1023/B:STCO.0000035301.49549.88
• Sofianos, E., Alexakis, C., and Gogas, P. (2025).Machine learning forecasting in the macroeconomic environment: the case of the US output gap. Econ Change Restruct ,58 (9), p.1-19. https://doi.org/10.1007/s10644-024-09849-w
• Sonmez, L. and Arslan, M. C. (2024). Volatility-Based Stock Market Forecasting with LSTM Model. International Journal of Accounting and Finance Researches, 6(2), p.48-61. https://dergipark.org.tr/en/download/article-file/4301777
• Sudiatmika, I. P. G. A., and Putra, I. M. A. W. (2024). Comparison of LSTM and GRU Models Performance in Forecasting Gold Prices: A Case Study Using Historical Data from Yahoo Finance. ARRUS Journal of Engineering and Technology, 4(1), p.157-165. https://doi.org/10.35877/jetech2760
• Tay, F. E. and Cao, L. (2001), Application of support vector machines in financial time series forecasting, Omega, 29(4), p.309-317. https://doi.org/10.1016/S0305-0483(01)00026-3
• Tenorio, J., and Pérez, W. (2024). Monthly GDP nowcasting with Machine Learning and Unstructured Data. Recuperado de, https://arxiv.org/abs/2402.04165
• Turan, T. and Afsal, M. Ş. (2020). The Determinants of Current Account Deficit in Turkey: An Empirical Anaylsis. Journal of Financial Politic & Economic Reviews, 651, p.217-236. https://www.ekonomikyorumlar.com.tr/files/articles/1584978012.pdf
• Yurtsever, M. (2023). Unemployment rate forecasting: LSTM-GRU hybrid approach. J Labour Market Res 57, 18. p.1-9. https://doi.org/10.1186/s12651-023-00345-8
• Yüksel, S. (2016). The Determinants of Current Account Deficit in Turkey: An Analysis with Mars Model. Bankers Journal, 96, p.102-121. https://www.tbb.org.tr/bankacilik/arastirma-ve-yayinlar/bankacilar-dergisi/pdf/809 Yıldız, Ş. (2020). The Sustainability of Turkey’s Current Accounts Deficit. Journal of Humanities and Tourism Research, 10(2), p.289-304. https://dergipark.org.tr/tr/download/article-file/1178258
• Zhou, Y., Li, T., Shi, J. ve Qian, Z. (2019). A CEEMDAN and XGBOOST-based approach to forecast crude oil prices. Complexity, 2019, p.1-15.  https://doi.org/10.1155/2019/4392785