Türkiye'de Tütün Piyasasının Yapay Sinir Ağları ile Tahmin Edilmesi

Yıl 2025, Cilt: 15 Sayı: 2, 688 - 701, 15.06.2025

Aysel Topşir , Ferdi Güler

https://doi.org/10.31466/kfbd.1570632

Öz

Bu çalışma, Türkiye'deki tütün politikalarının gelecekteki dinamiklerini yapay sinir ağları kullanarak öngörmeyi amaçlamaktadır. 1961–2022 yılları arasındaki tütün üretimi, hasat alanı ve verim verileri, bu değişkenler arasındaki karmaşık ilişkileri anlamak amacıyla kapsamlı bir şekilde analiz edilmiştir. Sonuçlar, 2023 ile 2027 yılları arasında tütün üretimi ve hasat alanının kademeli olarak azalmasının beklendiğini, buna karşılık verimin önemli ölçüde artacağını göstermektedir. Bu eğilim, teknolojik gelişmelerin ve etkili tarım politikalarının olumlu etkilerini yansıtmaktadır. Zaman serisi tahminleri derin dendritik yapay sinir ağları (DeepDenT) yazılımı kullanılarak gerçekleştirilmiştir. Bu tahminler, tütün tarımının sürdürülebilirliği ve stratejik planlaması açısından değerli bilgiler sunmaktadır. Tahminlerin yanı sıra, çalışmada değişkenler arasındaki ilişkileri değerlendirmek amacıyla doğrusal Granger nedensellik testi uygulanmıştır. Ancak, istatistiksel olarak anlamlı bir nedensellik bulunamamıştır; bu durum, tütün üretiminin karmaşık ve doğrusal olmayan dinamiklerden etkilendiğini göstermektedir. Bu da geleneksel doğrusal modellerin üretim sürecinin gerçek doğasını yeterince yansıtamayabileceğini ima etmektedir. Genel olarak, bu çalışma tütün tarımındaki uzun vadeli eğilimlere dair kritik bulgular sunmakta ve politika geliştirme süreçlerine katkı sağlamaktadır. Üreticilerin bilinçli ve stratejik kararlar almalarına destek olmakta ve sektörün sürdürülebilirliği ile ekonomik istikrarına ilişkin anlayışı derinleştirmektedir. Böylece, veri temelli yaklaşımlar ve ileri düzey modelleme teknikleriyle üretim süreçlerinin optimize edilmesine yönelik yeni bir bakış açısı sunmaktadır.

Anahtar Kelimeler

Öngörü , Tütün Analizi , Yapay Sinir Ağları

Forecasting the Tobacco Market in Türkiye with Artificial Neural Networks

Öz

This study aims to forecast the future dynamics of tobacco policies in Türkiye using artificial neural networks. Tobacco production, area harvested, and yield data from 1961 to 2022 were comprehensively analyzed to understand the complex relationships among these variables. The results indicate that, while tobacco production and harvested area are expected to decline gradually between 2023 and 2027, yield will significantly increase. This trend reflects the positive impact of technological advancements and effective agricultural policies. Time series forecasting was conducted using DeepDenT software. These forecasts provide valuable insights for the sustainability and strategic planning of tobacco farming. In addition to forecasting, the study applied the linear Granger causality test to assess relationships between the variables. However, no statistically significant causality was found, suggesting that tobacco production is influenced by complex, non-linear dynamics. This implies that conventional linear models may be insufficient to capture the true nature of the production process. Overall, the study offers critical insights into long-term trends in tobacco agriculture and contributes to policy development. It supports producers in making informed, strategic decisions and enhances understanding of the sector’s sustainability and economic stability. Thus, the study offers a new perspective on optimizing production through data-driven approaches and advanced modeling.

Anahtar Kelimeler

Forecasting , Analyzing Tobacco , Artificial Neural Networks

Kaynakça

• Agraz, M., Goksuluk, D., Zhang, P., Choi, B. R., Clements, R., Choudhary, G., and Karniadakis, G. (2024). ML-GAP: Machine Learning-Enhanced Genomic Analysis Pipeline with Autoencoders and Data Augmentation. Frontiers in Genetics, 15, 1442759.
• Agraz, M. (2024). Evaluating single multiplicative neuron models in physics-informed neural networks for differential equations. Scientific Reports, 14(1), 19073.
• Alizadeh, M. J., and Nourani, V. (2024). Multivariate GRU and LSTM models for wave forecasting and hindcasting in the southern Caspian Sea. Ocean Engineering, 298, 117193.
• Allende, H., Moraga, C., Salas, R. (2002). Artificial neural networks in time series forecasting: A comparative analysis. Kybernetika, 38(6), 685-707.
• Alpaslan, F., Egrioglu, E., Aladag, C. H., Tiring, E. (2012). Forecasting the lowest and highest gold prices using artificial neural networks. Journal of Statistical Research, 9(2), 12-19.
• Altan, S. (2008). An alternative approach for forecasting exchange rate performance using artificial neural networks. Journal of Gazi University Faculty of Economics and Administrative Sciences, 10(2), 141-160.
• Anonymous, (2021). Food and Agriculture Organization of the United Nations. Available from: http://www.fao.org/faostat/en/#data/QC.
• Anonymous, (2022). Turkish Statistical Institute. Available from: https://biruni.tuik.gov.tr/medas/?kn=92&locale=tr.
• Ataseven, B. (2013). Forecasting modeling with artificial neural networks. Journal of Proposal, 10(39), 101-115.
• Asi, E., Gozum, S. (2020). Reactions to policies on tobacco products and their implications in Turkey. Journal of Anatolian Nursing and Health Sciences, 23(2), 319-330.
• Bas, E., and Egrioglu, E. (2024, September 5–6). A new automatic forecasting method based on deep dendritic artificial neural network. The 18th International Days of Statistics and Economics, Prague, Czech Republic, pp. 24–31. https://msed.vse.cz/msed_2024/article/msed-2024-778-paper.pdf.
• Benli, Y. K., Yıldız, A. (2015). Forecasting gold prices using time series methods and artificial neural networks. Journal of Dumlupınar University Social Sciences, (42).
• Bilir, N., Cakir, B., Dagli, E., Erguder, T., Onder, Z. (2010). Tobacco control policies in Turkey. World Health Organization Report. Available from: URL: http://www.euro.who.int/documentE93038.
• Celik, S. (2020). Estimation modelling of tobacco production in Turkey: comparativ e analysis of artificial neural networks and multiplicative decomposition methods. International Journal of Trend in Research and Development, 7(4), 154-187.
• Chen, G., Jiang, S., Chen, D. (2024). Enhanced tobacco yield prediction using spatial information and exogenous variable-driven machine learning models. Journal of Scientific Research and Reports, 30(9), 2401.
• Egrioglu, E., and Bas, E. (2024). A new deep neural network for forecasting: Deep dendritic artificial neural network. Artificial Intelligence Review, 57(7), 171.
• Erilli, N. A., Egrioglu, E., Yolcu, U., Aladag, C. H., Uslu, V. R. (2010). Forecasting inflation in Turkey using a hybrid approach of feed-forward and feedback artificial neural networks. Journal of Dogus University, 11(1), 42-55.
• Guler, D., Saner, G., Naseri, Z. (2017). Forecasting the import volumes of oilseed crops using ARIMA and artificial neural network methods. Journal of Balkan and Near Eastern Social Sciences, 3(1), 60-70.
• Gumus, A. H., Gumus, S. G. (2005). Current practices and future of the tobacco sector in Turkey. Journal of Agricultural Economics, 11(1 and 2), 81-89.
• Guneri Tosunoglu, N., Keskin Benli, Y. (2012). Forecasting of Morgan Stanley Capital International Turkey Index using artificial neural networks. Ege Academic Review, 12(4).
• Guner, S. N., Demir, H. U. (2022). Forecasting iron and steel imports using artificial neural networks and time series methods. Journal of Sakarya Economics, 11(3), 389-397.
• Karabacak, K. (2017). Tobacco farming and its geographical distribution in Turkey. Journal of Geographical Sciences, 15(1), 27-48.
• Kaynar, O., Tastan, S. (2009). Comparison of MLP artificial neural networks and ARIMA models in time series analysis. Journal of Erciyes University Faculty of Economics and Administrative Sciences, (33).
• K. Sameshima and L. A. Baccala, Methods in Brain Connectivity Inference Through Multivariate Time Series Analysis. Boca Raton, FL, USA: CRC Press, 2016.
• Liu, X., and Wang, W. (2024). Deep time series forecasting models: A comprehensive survey. Mathematics, 12(10), 1504.
• Magazzino, C., Mele, M., and Schneider, N. (2021). A machine learning approach on the relationship among solar and wind energy production, coal consumption, GDP, and CO2 emissions. Renewable Energy, 167, 99-115.
• Mata, A. S. D. (2020). Complex networks: a mini-review. Brazilian Journal of Physics, 50, 658-672.
• Ozkan, F. (2011). An alternative approach to exchange rate forecasting with artificial neural networks. Journal of Eskisehir Osmangazi University, Faculty of Economics and Administrative Sciences, 2.
• Pesen, S. F., Karadogan, S., Akbulut, A. (2021). A general overview of tobacco use and tobacco control policies in the world and Turkey. Turkey Health Literacy Journal, 2(3), 191-196.
• Shi, X., Chen, Z., Wang, H., Yeung, D. Y., Wong, W. K., and Woo, W. C. (2015). Convolutional LSTM network: A machine learning approach for precipitation nowcasting. Advances in neural information processing systems, 28.
• Si, Y., Nadarajah, S., Zhang, Z., and Xu, C. (2024). Modeling opening price spread of Shanghai Composite Index based on ARIMA-GRU/LSTM hybrid model. PLOS ONE, 19(3), e0299164.
• Voyant, C., Notton, G., Kalogirou, S., Nivet, M. L., Paoli, C., Motte, F., and Fouilloy, A. (2017). Machine learning methods for solar radiation forecasting: A review. Renewable energy, 105, 569-582.
• Xiao, J., Bi, S., and Deng, T. (2024). Comparative analysis of LSTM, GRU, and Transformer models for stock price prediction. Manuscript submitted for publication.
• Zhang, M., Wang, J., Li, X. (2020). Tobacco yield prediction and disease detection using hyperspectral imaging and machine learning methods (BPNN, RF). Journal of Agricultural Engineering and Remote Sensing, 35(2), 123-135.
• Zorlutuna, S., Bircan, H. (2019). Comparison of time series analysis and artificial neural network methods in forecasting the number of tourists coming to Turkey. Journal of Cumhuriyet University Faculty of Economics and Administrative Sciences, 20(2), 164-185.