İşsizlik Oranı Öngörülerinde Makine Öğrenimi Yaklaşımları: Türkiye Üzerine Bir Uygulama

Yıl 2024, , 1 - 14, 28.06.2024

Polad Aliyev

https://doi.org/10.58618/igdiriibf.1477486

Öz

İşsizlik, sadece kapsamlı bir ekonomik sorun değil, aynı zamanda tüm ulusların odak noktası haline gelen karmaşık bir sosyal sorundur. İşsizlik sorununun doğru bir şekilde ele alınması, ülkenin kalkınmasıyla doğrudan ilişkilidir. Bu yönde oluşturulan politikaların başarası, işsizlik oranının doğru bir şekilde tahmin edilmesine dayanır. Bu makale, işsizlik oranı tahmininin yapılmasında yapay zekâ, makine öğrenimi ve klasik yöntemlerin kıyaslamasını amaçlamaktadır. Bu amaçla, Türkiye İstatistik Kurumu'ndan (TÜİK) Ocak 2005 verileriyle Aralık 2023 dönemini kapsayan işsizlik oranı verileri elde edilmiştir. Araştırmada ölçüt modeli olarak ARIMA, SARIMA modelleri, makine öğrenimi modelleri olarak Rassal Orman, XGBoost, LSTM ve GRU modelleri uygulanmıştır. Elde edilen sonuçlar, SARIMA'nın tahmin grafiğinin ve performans göstergelerinin ARIMA modeli performans değerlerinden daha iyi olduğunu göstermektedir. Makine öğrenimi modellerinde, MAPE dışındaki tüm hata ölçütleri SARIMA modelinin hata ölçütlerinden daha yüksektir. Ayrıca, bu modellerin belirleme katsayısı (R2) da SARIMA modelinin belirleme katsayısından (R2) daha büyüktür. Elde edilen sonuçlar en uygun metrik göstergeleri sergileyen makine öğrenimi yönteminin GRU modeli olduğunu ortaya koymuştur. Bu modelin MAE (Hataların Mutlak Değerlerinin Ortalaması) ve RMSE (Hata Kareler Ortalamasının Karekökü) değerleri en düşükken, R2 ise en yüksektir. Buna en yakın göstergeleri Rassal Orman modeli sergilemektedir.

Anahtar Kelimeler

Yapay Zekâ, İşsizlik Oranı, GRU, LSTM, Tek Değişkenli Zaman Serisi Modelleri

Machine Learning Approaches in Unemployment Rate Prediction: An Application on Türkiye

Öz

Unemployment is a complex economic and social issue affecting nations' development. Accurate unemployment rate estimation is crucial for successful policies aimed at addressing this issue. In this context, this article aims to compare artificial intelligence, machine learning, and classical methods of unemployment rate estimation. For this purpose, unemployment rate data covering the period between January 2005 and December 2023 were obtained from the Turkish Statistical Institute (TUİK). ARIMA and SARIMA models were applied as benchmark models, and Random Forest, XGBoost, LSTM, and GRU models were applied as machine learning models. The results show that the prediction graph and performance indicators of SARIMA are better than the ARIMA model performance values. In the machine learning models used in this study, all error measures except MAPE were higher than the error measures of the SARIMA model. Also, the coefficient of determination (R2) of these models was larger than that of the coefficient of determination of the SARIMA model. Furthermore, the results revealed that the machine learning method that exhibits the most favorable metric indicators is the GRU model. This model's MAE (Mean Absolute Value of Errors) and RMSE (Root Mean Square Error Squared) values were the lowest, while R2 was the highest. The Random Forest model exhibited the closest indicators.

Anahtar Kelimeler

Artificial Intelligence, Unemployment Rate, GRU, LSTM, Univariate Time Series Models

Kaynakça

• Abar, H. (2020). Xgboost ve Mars yöntemleriyle altın fiyatlarının kestirimi. EKEV Akademi Dergisi, 24(83), 427-446.
• Ahmad, M., Khan, Y. A., Jiang, C., Kazmi, S. J. H. ve Abbas, S. Z. (2021). The impact of COVID‐19 on unemployment rate: An intelligent based unemployment rate prediction in selected countries of Europe. International Journal of Finance & Economics, 28, 528–543 https://doi.org/10.1002/ijfe.2434
• Altındağ, İ. (2020). Karar ağacı ve rassal orman regresyon modeli. Veri madenciliğinde kullanılan regresyon modelleri ve R ile uygulamalı örnekler, Ö. Fruk Rençber (der.) içinde, 35-54.
• Arda, E. (2020). Yapay zekâ yöntemleri ile finansal zaman serisi öngörüleri (Yayımlanmamış doktora tezi). Başkent Üniversitesi Sosyal Bilimler Enstitüsü, Ankara.
• Arda, E. ve Küçükkocaoğlu, G. (2021). Yapay zekâ yöntemleri ile hisse senedi fiyat öngörüleri. Ekonomi, Politika ve Finans Araştırmaları Dergisi, 6(2), 565-586.
• Borkar, P. (2016). Modeling of groundnut production in India using ARIMA Model. International Journal of Research IT Management, 6(3), 36–44.
• Brownlee, J. (2018). XGBoost with Python. Machine Learning Mastery.
• Brownlee, J. (2020a). Time series forecasting with the Long Short-Term Memory Network in Python. Machine Learning Mastery. https://machinelearningmastery.com/time-series-forecasting-long-short-term-memory-network-python/
• Brownlee, J. (2020b). Random forest for time series forecasting. Machine Learning Mastery https://machinelearningmastery.com/random-forest-for-time-series-forecasting/
• Brownly, J. (2021). How to use XGBoost for time series forecasting. Machine Learning Mastery https://machinelearningmastery.com/xgboost-for-time-series-forecasting/
• Carmona, P., Climent, F. ve Momparler, A. (2019). Predicting failure in the U.S. banking sector: An extreme gradient boosting approach. International Review of Economics & Finance, 61, 304-323.
• Celbiş, M. G. (2022). Unemployment in rural Europe: A machine learning perspective. Applied Spatial Analysis and Policy, 16, 1071–1095. https://doi.org/10.1007/s12061-022-09464-0
• Çelik, Ş. (2019). Estimation of the orange production in Turkey by means of artificial neural networks. Global Journal of Engineering Science and Researches, 6(9), 10-16.
• Cerqueira, V., Torgo, L. ve Mozetič, I. (2020). Evaluating time series forecasting models: An empirical study on performance estimation methods. Machine Learning, 109, 1997–2028. https://doi.org/10.1007/s10994-020-05910-7
• Chen, C. (2006). Application of the novel nonlinear grey Bernoulli model for forecasting unemployment rate. Chaos, Solitons and Fractals, 37, 278–287.
• Çiftçi, S. ve Sir, B. G. D. (2023). Acil servise başvuru sayısının zaman serisi analiz ve makine öğrenmesi yöntemleri ile tahmin edilmesine yönelik bir uygulama. Pamukkale Üniversitesi Mühendislik Bilimleri Dergisi, 29(7), 667-679.
• Çöltekin, Ç. ve Rama, Ta. (2018). Tübingen-Oslo at SemEval-2018 Task 2: SVMs perform better than RNNs in Emoji Prediction. In Proceedings of the 12th International Workshop on Semantic Evaluation (ss. 34–38). Association for Computational Linguistics.
• Dzhunkeev, U. (2022). Forecasting unemployment in Russia using machine learning methods. Russian Journal of Money and Finance, 81(1), 73–87. Friedman, J. H. (2001). Greedy function approximation: A gradient boosting machine. Annals of Statistics, 29(5), 1189-1232.
• Gabrikova, B., Svabova, L. ve Kramarova, K. (2023). Machine learning ensemble modelling for predicting unemployment duration. Applied Sciences, (13), 10146. https://doi.org/10.3390/app131810146
• Graves, A. (2012). Supervised Sequence Labelling. Springer.
• Graves, A., Jaitly, N. ve Mohamed, A. R. (2013). Hybrid speech recognition with deep bidi-rectional LSTM. In Proceedings of the 2013 IEEE Workshop on Automatic Speech Recognition and Understanding (ASRU-2013) (ss. 273–278).
• Hatipoğlu, Ş., Belgrat, M. A., Degirmenci, A. ve Karal, Ö. (2021). Prediction of unemployment rates in Turkey by k-Nearest Neighbor regression analysis. In 2021 Innovations in Intelligent Systems and Applications Conference (ASYU) (ss. 1-5). IEEE.
• Hettiarachchi, H. ve Ranasinghe, T. (2019). Emoji powered capsule network to detect type and target of offensive posts in social media. In Proceedings of the International Conference on Recent Advances in Natural Language Processing (RANLP 2019) (ss. 474–480). INCOMA Ltd.
• Ho, T. (2022). Forecasting unemployment via machine learning: The use of average windows forecasts. SSRN. http://dx.doi.org/10.2139/ssrn.3496138
• Karahan, M. ve Çetintaş, F. (2022). Forecasting of Turkey's unemployment rate for future periods with artificial neural networks. Erciyes Üniversitesi İktisadi ve İdari Bilimler Fakültesi Dergisi, (62), 163-184. https://doi.org/10.18070/erciyesiibd.1056618
• Krollner, B., Vanstone, B. ve Finnie, G. (2010). Financial time series forecasting with machine learning techniques: A survey. Paper presented at the Proceedings of the 18th European Symposium on Artificial Neural Networks: Computational Intelligence and Machine Learning. https://pure.bond.edu.au/ws/files/27498056/Financial_time_series_forecasting_with_machine_learning_techniques.pdf
• Liu, X. ve Li, L. (2022). Prediction of labor unemployment based on time series model and neural network model. Hindawi, Computational Intelligence and Neuroscience. https://doi.org/10.1155/2022/7019078
• Li, X. ve Yang, T. (2021). Forecast of the employment situation of college graduates based on the LSTM neural network. Hindawi Computational Intelligence and Neuroscience, 1-11. https://doi.org/10.1155/2021/5787355
• Mulaudzi, R. ve Ajoodha, R. (2020). An exploration of machine learning models to forecast the unemployment rate of South Africa: A univariate approach. 2nd International Multidisciplinary Information Technology and Engineering Conference (IMITEC), 25-27 Nov. Kimberley, Güney Afrika.
• Mutascu, M. (2021). Artificial intelligence and unemployment: New insights. Economic Analysis and Policy, 69, 653-667.
• Ma, X., Sha, J., Wang, D., Yu, Y., Yang, Q. ve 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, 24-39.
• Olmedo, E. (2014). Forecasting Spanish unemployment using near neighbour and neural net techniques. Computational Economics, 43, 183–197. https://doi.org/10.1007/s10614-013-9371-1.
• Raşo, H., ve Demirci, M. (2019). Predicting the Turkish stock market BIST 30 index using deep learning. International Journal of Engineering Research and Development, 11(1), 253-265.
• Sen, M., Basu, S., Chatterjee, A., Banerjee, A., Pali, S. P. K., ve Mukhopadhyay, Dutta, S. ve Tarafdar, A. (2022). Prediction of unemployment using machine learning approach. In 2022 OITS International Conference on Information Technology (OCIT) (ss. 1-5). Bhubaneswar, Hindistan. https://doi.org/10.1109/OCIT56763.2022.00072
• Shen, S., Jiang, H. ve Zhang, T. (2012). Stock market forecasting using machine learning algorithms (Stanford University Working Paper). http://cs229.stanford.edu/proj2012/ShenJiangZhang-StockMarketForecastingusingMachineLearningAlgorithms.pdf
• Stasinakis, C., Sermpinis, G., Theofilatos, K. ve Karathanasopoulos, A. (2014). Forecasting US unemployment with radial basis neural networks, Kalman filters and support vector regressions. Computational Economics, 47, 569–587. https://doi.org/10.1007/s10614-014-9479-y
• Tsai, C.-F. ve Wang, S.-P. (2009). Stock price forecasting by hybrid machine learning techniques. Paper presented at the International MultiConference of Engineers and Computer Scientists. http://www.iaeng.org/
• Van den Berg, G. J., Kunaschk, M., Lang, J., Stephan, G. ve Uhlendorff, A. (2023). Predicting re-employment: Machine learning versus assessments by unemployed workers and by their caseworkers. [IZA DP No. 16426]. IZA Institute of Labor Economics.
• Yamaclı, S. ve Yamaclı, S. (2023). Estimation of the unemployment rate in Turkey: A comparison of the ARIMA and machine learning models including COVID-19 pandemic periods. Heliyon, 9(1), e12796. https://doi.org/10.1016/j.heliyon.2023.e12796
• Yang, S., Yu, X. ve Zhou, Y. (2020). LSTM and GRU neural network performance comparison study: Taking Yelp review dataset as an example. In 2020 International Workshop on Electronic Communication and Artificial Intelligence (IWECAI) (ss. 98-101). Shanghai, Çin. https://doi.org/10.1109/IWECAI50956.2020.00027
• Yıldırım, H. ve Başeğmez, H. (2017). Analysis and forecast of Turkey unemployment rate. Global Journal of Mathematical Analysis, 5(1), 11-15. https://doi.org/10.14419/gjma.v5i1.6841.
• Yurtsever, M. (2023). Unemployment rate forecasting: LSTM GRU hybrid approach. Journal for Labour Market Research, 57(18). https://doi.org/10.1186/s12651-023-00345-8