Performance enhancements of classification models using the undersampling method (NearMiss).

Performance enhancements of classification models using the undersampling method (NearMiss).

<p>(a) AUC improvement, (b) MCC improvement, and (c) F1 score improvement achieved by applying NearMiss in comparison to the original GMSC dataset across each model.</p>

محفوظ في:

التفاصيل البيبلوغرافية
المؤلف الرئيسي:	Seongil Han (20490232) (author)
مؤلفون آخرون:	Haemin Jung (20490235) (author)
منشور في:	2024
الموضوعات:	Science Policy Biological Sciences not elsewhere classified Mathematical Sciences not elsewhere classified Information Systems not elsewhere classified understanding individual predictions providing detailed insights model &# 8217 managing class imbalance majority class tends imbalanced data distributions imbalanced class distributions explainability aspects within experimental results show evaluating borrower risk achieving optimal metrics credit risk classification predictive power decreases enhance model performance combining oversampling techniques integrates interpretability features enhancing model interpretability explainable credit scoring credit scoring due improving predictive performance parametric oversampling approach nate enhances interpretability credit scoring predictive performance parametric approach oversampling approaches transparent tool sustaining profitability reduced accuracy paper introduces nate ), making process logistic regression linear datasets gradient boosting framework designed financial institutions feature contributions f1 score crucial role computational efficiency comprehensively addresses based classifiers
الوسوم:	إضافة وسم لا توجد وسوم, كن أول من يضع وسما على هذه التسجيلة!

Cannot write session to /tmp/vufind_sessions/sess_4r8lrfu27ce2hu6mhpdad83jgp