Gradient Boosting (XGBoost) bilan kredit riskini baholash
Keywords:
Kredit riski, XGBoost, Gradient Boosting, mashinaviy o‘rganish, PD modeli, defolt ehtimoli, SHAP qiymatlari, ansambl modellar, ma’lumotlarni tayyorlash, klassifikatsiyaAbstract
Ushbu ilmiy maqolada kredit riskini baholashda Gradient Boosting oilasiga mansub XGBoost algoritmining qo‘llanishi chuqur tahlil qilinadi. Tadqiqotda ma’lumotlar to‘plamining xususiyatlari, ularning tozalanishi va modellashtirishga tayyorlanishi, XGBoost algoritmining matematik asoslari, gipertuzilmalarning ta’siri va modellarni baholash indikatorlari keng yoritiladi. Amaliy natijalarda modelning yuqori ajratuvchanlik qobiliyati, defolt ehtimolini prognozlashdagi ustunliklari hamda SHAP orqali interpretatsiya qilinishi ko‘rsatib berilgan. Tadqiqot kredit riskini sakkizlik yondashuvlarda avtomatlashtirish, risklarni boshqarish tizimini raqamlashtirish va portfel barqarorligini oshirishda XGBoostning amaliy ahamiyatini tasdiqlaydi.
References
1. Chen, T., & Guestrin, C. (2016). “XGBoost: A Scalable Tree Boosting System.” KDD Conference.
2. Friedman, J. H. (2001). “Greedy Function Approximation: A Gradient Boosting Machine.”
3. Brown, I., & Mues, C. (2012). “Classification Algorithms for Imbalanced Credit Scoring Data.”
4. Hand, D., & Henley, W. (1997). “Statistical Classification in Consumer Credit.”
5. Shapley, L. (1953). “A Value for n-Person Games.”
6. Bishop, C. (2006). Pattern Recognition and Machine Learning.
7. Hastie, Tibshirani & Friedman (2009). Elements of Statistical Learning.
8. Basel Committee on Banking Supervision (BCBS) materials.
