Çoklu Doğrusal Bağlantılı Nadir Olayların Modellenmesinde Lasso ve Ridge Regresyon ile Boosting Algoritmalarının Performans Karşılaştırması

Yıl 2024, Cilt: 9 Sayı: 1, 154 - 166, 29.06.2024

Olcay Alpay

https://doi.org/10.33484/sinopfbd.1434260

Öz

Bu çalışma, iki durumlu olayları modellemek için kullanılan makine öğrenmesi tekniklerinde karşılaşılan nadirlik ve “çoklu doğrusal bağlantı” ya da sadece “çoklu bağlantı” olarak tanımlanan sorunu ele alınmaktadır. Çoklu doğrusal bağlantı (ÇDB), bağımsız değişkenler arasında bir ya da birden fazla kuvvetli doğrusal bağımlılık olma durumudur ve bir sorun olarak ortaya çıkar. Üzerinde çalışılan veri içerisinde çoklu doğrusal bağlantı probleminin var olması regresyon katsayılarının varyanslarının büyümesi gibi olumsuz bir sonuca sebebiyet verir. Bu çalışmada, Lasso ve Ridge Regresyon ile GradientBoost, XGBoost, LightGBM ve AdaBoost gibi artırma algoritmaları içeren düzenleme ve ölçeklendirme tekniklerinin, çoklu doğrusal bağlantılı nadir olayların modellenmesinde, algoritmaların performanslarını karşılaştırmak için detaylı bir simülasyon çalışması sunulmaktadır. Simülasyon çalışmasında, verideki dengesizliği ortadan kaldırmak amacıyla yeniden örnekleme yöntemleri kullanılarak sonuçlara etkisi Hata Kareler Ortalaması (HKO), R^2, Hassasiyet (Precision-Prec), Duyarlılık (Recall-Rec) ve Eğri Altında Kalan Alan (Area Under the Curve-AUC) gibi performans metrikleri ve İşlem Karakteristik Eğrisi (Receiver Operating Characteristic- ROC) grafikleri ile araştırılmaktadır. Sonuçlar Lasso, Ridge ve Boosting algoritmalarının ÇDB’ya sahip nadir olayların modellenmesinde hangi yöntemin uygun olduğunu belirlemek açısından katkı sunmaktadır.

Anahtar Kelimeler

Lasso regresyon, Ridge regresyon, Boosting algoritmaları, performans metrikleri, yeniden örnekleme teknikleri

Performance Comparison of Lasso and Ridge Regression and Boosting Algorithms for Modeling Rare Events with Multicollinearity

Öz

This study examines the issues of rarity and multicollinearity in machine learning techniques used to model binary events. Multicollinearity (MC) is the presence of strong linear dependencies among independent variables, which poses a problem. In the context of the data being studied, the existence of multicollinearity leads to undesired consequences such as an enlargement of the variances of the regression coefficients. This study presents a simulation comparing the performance of algorithms in modelling multicollinear and rare events. Regularization and scaling techniques such as Lasso and Ridge Regression, as well as Boosting algorithms like GradientBoost, XGBoost, LightGBM, and AdaBoost are utilized. The impact of resampling methods to reduce data imbalance is also investigated using performance metrics such as Mean Squared Error (MSE), R^2, Precision (Prec), Recall (Rec) and AUC values, along with ROC curves. The results help to determine the appropriate method for modelling rare events with multicollinearity and provide insight into the performance of Lasso, Ridge and Boosting algorithms.

Anahtar Kelimeler

Lasso regression, Ridge regression, Boosting algorithms, performance metrics, resampling techniques

Kaynakça

• Bayman, O. E., & Dexter, F. (2021). Multicollinearity in logistic regression models. Anesthesi̇a & Analgesi̇a, 133(2), 362-365. https://doi: 10.1213/ane.0000000000005593
• King, G., & Zeng, L. (2001). Logistic regression in rare events data. Poli̇ti̇cal Analysi̇s, 9(2), 137-163. https://doi:10.1093/oxfordjournals.pan.a004868
• Maalouf, M., & Trafalis, T. B. (2011). Robust weighted kernel logistic regression in imbalanced and rare events data. Computati̇onal Stati̇sti̇cs & Data Analysi̇s, 55(1), 168-183. https://doi:10.1016/j.csda.2010.06.014
• Shrivastava, S., Jeyanthi, P. M., & Singh, S. (2020). Failure prediction of Indian Banks using SMOTE, Lasso regression, bagging and Boosting. Cogent Economics & Finance, 8(1), 1729569. https://doi.org/10.1080/23322039.2020.1729569
• Rochayani, M. Y., Sa'adah, U., & Astuti, A. B. (2020). Finding biomarkers from a high-dimensional imbalanced dataset using the hybrid method of random undersampling and lasso. Comtech: Computer, Mathemati̇cs and Engi̇neeri̇ng Appli̇cati̇ons, 11(2), 75-81. https://doi:10.21512/comtech.v11i2.6452
• Cahyana, N., Khomsah, S., & Aribowo, A. S. (2019). Improving imbalanced dataset classification using oversampling and gradient Boosting [Bildiri sunumu]. 5th international conference on science in information technology (ICSITech), China.
• Tanha, J., Abdi, Y., Samadi, N., Razzaghi, N., & Asadpour, M. (2020). Boosting methods for multi-class imbalanced data classification: an experimental review. Journal of Big Data, 7, 1-47. https://doi.org/10.1186/s40537‑020‑00349‑y
• Ashraf, M. T., Dey, K., & Mishra, S. (2023). Identification of high-risk roadway segments for wrong-way driving crash using rare event modeling and data augmentation techniques. Accident Analysis & Prevention, 181, 106933. https://doi.org/10.1016/j.aap.2022.106933
• Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of the Royal Statistical Society Series B: Statistical Methodology, 58(1), 267-288.
• Göv, A., & Kapkara Kaya, S. (2023). Türkiye örneğinde çevresel kalitenin belirleyicileri: lasso yaklaşımı. Pamukkale Üniversitesi Sosyal Bilimler Enstitüsü Dergisi, (54), 25-37. https://doi:10.30794/pausbed.1097352
• Yüzbaşı, B., & Pala, M. (2022). Ridge regresyon parametre seçimi: Türkiye’nin doğrudan yabancı yatırım örneği. İstatistikçiler Dergisi: İstatistik ve Aktüerya, 15(1), 1-18.
• Mahesh, B. (2020). Machine learning algorithms - a review. International Journal of Science and Research (IJSR), 9(1), 381-386. https://doi: 10.21275/ART20203995
• Friedman, J. H. (2002). Stochastic gradient Boosting. Computational Statistics & Data Analysis, 38(4), 367-378.
• Ali, Z. A., Abduljabbar, Z. H., Taher, H. A., Sallow, A. B., & Almufti, S. M. (2023). Exploring the power of extreme gradient Boosting algorithm in machine learning: A review. Academic Journal of Nawroz University, 12(2), 320-334.
• Tyralis, H., & Papacharalampous, G. (2021). Boosting algorithms in energy research: A systematic review. Neural Computing and Applications, 33(21), 14101-14117. https://doi.org/10.1007/s00521-021-05995-8
• Li, S., & Zhang, X. (2020). Research on orthopedic auxiliary classification and prediction model based on XGBoost algorithm. Neural Computing and Applications, 32(7), 1971-1979. https://doi.org/10.1007/s00521-019-04378-4
• Wang, D. N., Li, L., & Zhao, D. (2022). Corporate finance risk prediction based on LightGBM. Information Sciences, 602, 259-268. https://doi.org/10.1016/j.ins.2022.04.058
• Gu, Q., Sun, W., Li, X., Jiang, S., & Tian, J. (2023). A new ensemble classification approach based on Rotation Forest and LightGBM. Neural Computing and Applications, 35(15), 11287-11308. https://doi.org/10.1007/s00521-023-08297-3
• Ying, C., Qi-Guang, M., Jia-Chen, L., ve Lin, G. (2013). Advance and prospects of AdaBoost algorithm. Acta Automatica Sinica, 39(6), 745-758. https://doi.org/10.1016/S1874-1029(13)60052-X
• Hoens, T. R., & Chawla, N. V. (2013). Imbalanced datasets: from sampling to classifiers. H. He & Y. Ma (Ed.), Imbalanced learning: Foundations, algorithms, and applications, (s.43-59). Wi̇ley Onli̇ne Li̇brary.
• Wang, J., Xu, M., Wang, H., & Zhang, J. (2006). Classification of imbalanced data by using the SMOTE algorithm and locally linear embedding [Bildiri sunumu]. In 8th international conference on signal processing (IEEE), China.
• Birla, S., Kohli, K., & Dutta, A. (2016). Machine learning on imbalanced data in credit risk [Bildiri sunumu]. In 2016 IEEE 7th annual information technology, electronics and mobile communication conference (IEMCON), Canada.
• Wang, Z. H. E., Wu, C., Zheng, K., Niu, X., & Wang, X. (2019). SMOTETomek-based resampling for personality recognition. IEEE access, 7, 129678-129689. https://doi:10.1109/ACCESS.2019.2940061
• Werner de Vargas, V., Schneider Aranda, J. A., dos Santos Costa, R., da Silva Pereira, P. R., & Victória Barbosa, J. L. (2023). Imbalanced data preprocessing techniques for machine learning: a systematic mapping study. Knowledge and Information Systems, 65(1), 31-57. https://doi.org/10.1007/s10115-022-01772-8
• Mandrekar, J. N. (2010). Receiver operating characteristic curve in diagnostic test assessment. Journal of Thoracic Oncology, 5(9), 1315-1316. https://doi.org/10.1097/JTO.0b013e3181ec173d
• Keçeoğlu, Ç. R., Gelbal, S., & Doğan, N. (2016). Roc eğrisi yöntemi ile kesme puaninin belirlenmesi. The Journal of Academic Social Science Studies, 50(2), 553-562. http://dx.doi.org/10.9761/JASSS3564
• Oommen, T., Baise, L. G., & Vogel, R. M. (2011). Sampling bias and class imbalance in maximum-likelihood logistic regression. Mathematical Geosciences, 43, 99-120. https://doi10.1007/s11004-010-9311-8
• Huang, J., & Ling, C. X. (2005). Using AUC and accuracy in evaluating learning algorithms. IEEE Transactions on knowledge and Data Engineering, 17(3), 299-310.
• LemaÃŽtre, G., Nogueira, F., & Aridas, C. K. (2017). Imbalanced-learn: A python toolbox to tackle the curse of imbalanced datasets in machine learning. Journal of Machine Learning Research, 18(17), 1-5.