Journal of Advances in Developmental Research

E-ISSN: 0976-4844   •   Impact Factor: 9.71

A Widely Indexed Open Access Peer Reviewed Multidisciplinary Bi-monthly Scholarly International Journal

Call for Paper Volume 17 Issue 2 July-December 2026 Submit your research before last 3 days of December to publish your research paper in the issue of July-December.
Submit Research Paper Join as a Reviewer

Plagiarism is checked by the leading plagiarism checker TurnItIn

Call for Paper

Volume 17 Issue 2
2026

Submit your research paper

Indexing Partners
Academia Advanced Sciences Index Bielefeld Academic Search Engine CiteSeer DRJI Google Scholar Independent Search Engine & Directory Network ISI (International Scientific Indexing) Issuu Mendeley Research Networks
RefSeek ResearcherId - Thomson Reuters ResearchGate Scirus Scribd Semantic Scholar UTeM - Universiti Teknikal Malaysia Melaka Wiki for Call for Papers WorldCat Zenodo

Operationalizing Imbalance: Boundary vs. Density-Based Sampling in Extreme and Moderate Credit Risk Scenarios

Author(s) Sai Prashanth Pathi
Country United States
Abstract Machine learning models in credit risk are frequently compromised by the class imbalance problem, where fraud cases or defaults represent a negligible fraction of the population. While the Synthetic Minority Over-sampling Technique (SMOTE) is the de facto standard for addressing this, recent literature suggests it may introduce noise and computational overhead without operational gain. This study conducts a rigorous comparative analysis of six strategies: Cost-Sensitive Learning (Baseline), Random Undersampling (RUS), Vanilla SMOTE, Borderline-SMOTE, ADASYN, and SMOTE-ENN applied to Gradient Boosted Decision Trees (XGBoost). The evaluation utilizes three datasets with varying imbalance ratios (from 0.17% to 6.9%) to test robustness across different financial contexts. Our experiments reveal two critical insights. First, we identify a "False Positive Trap" in extreme imbalance scenarios: while Vanilla SMOTE achieved the highest Area Under Precision-Recall Curve (AUPRC: 0.825), it degraded the F1-score to 0.282, rendering it operationally inviable. In contrast, Borderline-SMOTE maintained a comparable AUPRC (0.818) while achieving a superior F1-score (0.690). Second, we detect a "Threshold of Necessity": in scenarios with moderate imbalance (>5%), all sampling techniques failed to outperform the cost-sensitive baseline, suggesting that synthetic sampling is counter-productive when sufficient minority examples exist.
Keywords Credit Risk, Class Imbalance, Fraud Detection, XGBoost, SMOTE, Anomaly Detection.
Field Engineering
Published In Volume 16, Issue 2, July-December 2025
Published On 2025-12-12
DOI https://doi.org/10.71097/IJAIDR.v16.i2.1691
Short DOI https://doi.org/hbphzs
View / Download PDF File

Share this



CrossRef

CrossRef DOI is assigned to each research paper published in our journal.

IJAIDR DOI prefix is
10.71097/IJAIDR

Downloads
Research Paper Format Copyright Permission Form and Undertaking Form Cover Page Vol 17 Isu 1Cover Page Vol 16 Isu 2Cover Page Vol 16 Isu 1

All research papers published on this website are licensed under Creative Commons Attribution-ShareAlike 4.0 International License, and all rights belong to their respective authors/researchers.

CC-BY-SA Open Access

About IJAIDR
Fees & Payment
Current Issue
Publication Archive 		Submit Research Paper
Track Submission Status
Publication Guidelines
Peer Review & Plagiarism 		Join as a Reviewer
Editors & Reviewers
Get Reviewer Membership Certi. 		Website/Journal Policies
Usage Policy
Content Policies
Privacy Policy
Contact Us Message on WhatsApp +91-9687-182-185 editor@ijaidr.com