SavedAdd your journal

Journals / Sakarya Üniversitesi İşletme Enstitüsü Dergisi / 2026 / Cilt: 8 Sayı: 1

Inflation Forecasting with LSTM Model in the Global Tension Environment: The Case of Türkiye

Küresel Gerginlik Ortamında LSTM Modeli ile Enflasyon Tahminlemesi: Türkiye Örneği

Nilay Özdemir(İZMİR KATİP ÇELEBİ ÜNİVERSİTESİ, SOSYAL BİLİMLER ENSTİTÜSÜ)0009-0001-5021-685X
Emrah Noyan(İZMİR KATİP ÇELEBİ ÜNİVERSİTESİ, İKTİSADİ VE İDARİ BİLİMLER FAKÜLTESİ)0000-0002-4482-0110
Pages
36–39
Publication Date
2026-06-30
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

The results indicate that external factors such as geopolitical vulnerabilities, exchange rate volatility, and energy price shocks must be integrated as exogenous features into the LSTM architecture.

📌 3 Key Takeaways

🎯
Objective & Hypothesis:This study aims to forecast the monthly Consumer Price Index (CPI) rates of Türkiye for the period 2005–April 2026 using the Long Short-Term Memory (LSTM) deep learning architecture.
📊
Key Findings & Data:The primary motivation stems from the impact of global geopolitical tensions—particularly the Israel-Iran conflict and the escalating Middle East crisis driven by US regional policies—on oil markets, and the transmission of this pressure to Turkish inflation through the exchange rate channel.
💡
Conclusion & Impact:The results indicate that external factors such as geopolitical vulnerabilities, exchange rate volatility, and energy price shocks must be integrated as exogenous features into the LSTM architecture.
🔬Design:Modeling & Simulation
📍Scope:2005 Verileri
⚙️Variables:Uzun Kısa Süreli Bellek (LSTM) · Enflasyon Tahmini · Derin Öğrenme

Abstract

This study aims to forecast the monthly Consumer Price Index (CPI) rates of Türkiye for the period 2005–April 2026 using the Long Short-Term Memory (LSTM) deep learning architecture. The primary motivation stems from the impact of global geopolitical tensions—particularly the Israel-Iran conflict and the escalating Middle East crisis driven by US regional policies—on oil markets, and the transmission of this pressure to Turkish inflation through the exchange rate channel. The LSTM model, applied to a dataset of 254 monthly observations, was configured with 24 hidden units, a lookback window of 20 steps, and the Huber loss function. The training loss declined to 0.0748, while the validation loss remained at 1.1787. For the test period, the computed RMSE was 2.80 percentage points and MAE was 2.22 percentage points; the R² value yielded a negative result of -1.0039. This negative R² signals that the model fell short in capturing the high volatility and structural breaks of the test period, thereby providing significant findings regarding the non-linear and regime-switching nature of Turkish inflation. The results indicate that external factors such as geopolitical vulnerabilities, exchange rate volatility, and energy price shocks must be integrated as exogenous features into the LSTM architecture.

Özet

Bu çalışma, Türkiye ekonomisinin 2005-2026 Nisan dönemi aylık Tüketici Fiyat Endeksi (TÜFE) oranlarını Long Short-Term Memory (LSTM) derin öğrenme mimarisi ile tahmin etmeyi amaçlamaktadır. Küresel jeopolitik gerginliklerin —özellikle İsrail-İran çatışması ve ABD'nin bölgesel politikaları çerçevesinde tırmanan Orta Doğu krizinin— petrol piyasaları üzerindeki baskısı ve bu baskının döviz kanalıyla Türkiye enflasyonuna yansıması temel motivasyon kaynağını oluşturmaktadır. 254 aylık gözlemi kapsayan veri seti üzerinde uygulanan LSTM modeli; 24 gizli birim, 20 adım geriye dönük pencere ve Huber kayıp fonksiyonu parametreleriyle yapılandırılmıştır. Eğitim verisi üzerinde elde edilen kayıp değeri 0,0748'e gerilemiş, ancak doğrulama kaybı 1,1787 düzeyinde kalmıştır. Test dönemi için hesaplanan RMSE 2,80 puan ve MAE 2,22 puan olarak gerçekleşmiş; R² değeri -1,0039 ile negatif bir sonuç vermiştir. Bu negatif R² değeri, modelin test periyodundaki yüksek oynaklık ve yapısal kırılmalar karşısında yetersiz kaldığına işaret etmekte olup söz konusu durum, Türkiye enflasyonunun doğrusal olmayan ve rejim-değişimli yapısına ilişkin önemli bulgular sunmaktadır. Sonuçlar, jeopolitik kırılganlıklar, döviz kuru oynaklığı ve enerji fiyatı şokları gibi dışsal faktörlerin LSTM mimarisine dışsal değişken (exogenous feature) olarak entegre edilmesi gerektiğini ortaya koymaktadır.

🔖

Cite This Article

Özdemir, N., Noyan, E. (2026). Inflation Forecasting with LSTM Model in the Global Tension Environment: The Case of Türkiye. Sakarya Üniversitesi İşletme Enstitüsü Dergisi, Cilt: 8 Sayı: 1, 36–39. https://doi.org/10.47542/sauied.1945798

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (21)

  1. Akçelik, F. ve Yücel, E. (2022). Türkiye'de döviz kuru geçiş etkisi ve enflasyon dinamikleri. İktisat ve Politika Dergisi, 12(1), 1-28.
  2. Arslanalp, S., Boz, E. ve García-Macia, D. (2019). Inflation components and monetary policy transmission in Turkey (IMF Working Paper No. WP/19/178). Uluslararası Para Fonu.
  3. Baek, E. ve Koo, W. W. (2010). An exchange rate model based on demand for imports and the balance of trade. Asian Economic Journal, 24(3), 215-234.
  4. Bengio, Y., Courville, A. ve Vincent, P. (2013). Representation learning: A review and new perspectives. IEEE Transactions on Pattern Analysis and Machine Intelligence, 35(8), 1798-1828. https://doi.org/10.1109/TPAMI.2013.50
  5. Caldara, D. ve Iacoviello, M. (2022). Measuring geopolitical risk. American Economic Review, 112(4), 1194-1225. https://doi.org/10.1257/aer.20191823
  6. Fisher, I. (1911). The purchasing power of money: Its determination and relation to credit interest and crises. Macmillan.
  7. Gordon, R. J. (1990). The measurement of durable goods prices. University of Chicago Press.
  8. Hamilton, J. D. (2003). What is an oil shock? Journal of Econometrics, 113(2), 363-398. https://doi.org/10.1016/S0304-4076(02)00207-5
  9. Hochreiter, S. ve Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8), 1735-1780. https://doi.org/10.1162/neco.1997.9.8.1735
  10. Kaya, H., ve Erdoğan, S. (2024). Oil price dynamics and inflation transmission in Turkey: A DCC-GARCH analysis. Energy Economics Letters, 11(2), 1-15.
  11. Keynes, J. M. (1936). The general theory of employment, interest and money. Macmillan.
  12. Kilian, L. (2009). Not all oil price shocks are alike: Disentangling demand and supply shocks in the crude oil market. American Economic Review, 99(3), 1053-1069. https://doi.org/10.1257/aer.99.3.1053
  13. Kingma, D. P. ve Ba, J. (2015). Adam: A method for stochastic optimization. Proceedings of the 3rd International Conference on Learning Representations (ICLR 2015), 1-15. https://arxiv.org/abs/1412.6980
  14. Lim, B. ve Zohren, S. (2021). Time-series forecasting with deep learning: A survey. Philosophical Transactions of the Royal Society A, 379(2194), 1-21. https://doi.org/10.1098/rsta.2020.0209
  15. Makridakis, S., Spiliotis, E. ve Assimakopoulos, V. (2018). Statistical and machine learning forecasting methods: Concerns and ways forward. PLOS ONE, 13(3), e0194889. https://doi.org/10.1371/journal.pone.0194889
  16. Medeiros, M. C., Vasconcelos, G. F. R. Veiga, Á., ve Zilberman, E. (2021). Forecasting inflation in a data-rich environment: The benefits of machine learning methods. Journal of Business and Economic Statistics, 39(1), 98-119. https://doi.org/10.1080/07350015.2019.1637745
  17. Nakagawa, K., Imakubo, K. ve Hanada, E. (2018). Predicting core consumer prices using machine learning. Bank of Japan Working Paper Series, E-4, 1-35.
  18. Siami-Namini, S. ve Siami-Namini, A. (2018). A comparison of ARIMA and LSTM in forecasting time series. Proceedings of the 2018 17th IEEE International Conference on Machine Learning and Applications (ICMLA), 1394-1401. https://doi.org/10.1109/ICMLA.2018.00227
  19. Türkiye İstatistik Kurumu [TÜİK]. (2026). Tüketici fiyat endeksi verileri, Ocak 2005 - Nisan 2026. https://data.tuik.gov.tr
  20. Yılmaz, M. ve Yurdakul, F. (2020). Machine learning approaches to forecasting inflation in Turkey. Sosyoekonomi, 28(43), 33-56.
  21. Zhang, G. P. (2003). Time series forecasting using a hybrid ARIMA and neural network model. Neurocomputing, 50, 159-175. https://doi.org/10.1016/S0925-2312(01)00702-0