How hmnthy/credit-risk-scoring-hmeq works
This is a credit-risk scoring project delivered primarily as a Google Colab notebook. Its main workflow takes HMEQ loan data through exploration and model selection, compares logistic regression with boosted-tree models, evaluates default predictions and decision thresholds, then produces a business-profile score grid. The README also identifies a profiling report and exported grid as deliverables. Notebook stages are represented separately but share the notebook path; the source excerpts contain no notebook code, so relationships are based on documented workflow statements rather than inspected calls.
Main entry points: Scoring.ipynb Model Preselection · grille_de_score_weight.xlsx Score Grid · hmeq_report.html Data Profiling · Rapport_Scoring.docx Project Report
Components 12
Data Analysis
- Exploratory Analysis
Scoring.ipynb - Data Profiling
hmeq_report.html
Risk Modeling
- Model Preselection
Scoring.ipynb - Logistic Regression
Scoring.ipynb - Gradient Boosting
Scoring.ipynb - Model Evaluation
Scoring.ipynb - Threshold Analysis
Scoring.ipynb
Risk Deliverables
- Risk Segmentation
Scoring.ipynb - Score Grid
grille_de_score_weight.xlsx - Project Report
Rapport_Scoring.docx
Other components
- HMEQ Dataset
- Credit Team
Connections 12
- HMEQ Dataset → Exploratory Analysis: loads data
README.md - HMEQ Dataset → Data Profiling: profiles data
README.md - Exploratory Analysis → Model Preselection: informs selection
README.md - Model Preselection → Logistic Regression: compares models
README.md - Model Preselection → Gradient Boosting: compares models
README.md - Logistic Regression → Model Evaluation: evaluated on test
README.md - Gradient Boosting → Model Evaluation: evaluated on test
README.md - Model Evaluation → Threshold Analysis: sets operating point
README.md - HMEQ Dataset → Risk Segmentation: segments training data
README.md - Risk Segmentation → Score Grid: exports grid
README.md - Credit Team → Score Grid: uses for policy
- Credit Team → Project Report: reviews results