https://github.com/fbarffmann/credit-risk-classification
Classified 19,000+ loans as high-risk or healthy using logistic regression. Achieved 100% precision for healthy loans and 84% precision for high-risk loans.
https://github.com/fbarffmann/credit-risk-classification
classification credit-risk data-analysis logistic-regression machine-learning model-evaluation pandas python scikit-learn
Last synced: about 22 hours ago
JSON representation
Classified 19,000+ loans as high-risk or healthy using logistic regression. Achieved 100% precision for healthy loans and 84% precision for high-risk loans.
- Host: GitHub
- URL: https://github.com/fbarffmann/credit-risk-classification
- Owner: fbarffmann
- Created: 2024-09-14T15:05:38.000Z (7 months ago)
- Default Branch: main
- Last Pushed: 2025-04-13T16:30:32.000Z (11 days ago)
- Last Synced: 2025-04-13T17:38:16.047Z (11 days ago)
- Topics: classification, credit-risk, data-analysis, logistic-regression, machine-learning, model-evaluation, pandas, python, scikit-learn
- Language: Jupyter Notebook
- Homepage:
- Size: 526 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Credit Risk Classification using Logistic Regression
Built and evaluated a machine learning model to classify loans as high-risk or healthy based on borrower data. Used logistic regression to predict loan status with a focus on precision and recall for high-risk loans.
## Tools & Technologies Used
- Python
- Pandas
- Scikit-learn (Logistic Regression)
- Classification Report
- Confusion Matrix
- Jupyter Notebooks
## File Structure
```text
.
├── Credit_Risk/
│ ├── credit_risk_classification.ipynb # Full model workflow
│ └── lending_data.csv # Loan dataset
```
## Skills Demonstrated
- Binary classification using Logistic Regression
- Data preparation and feature engineering
- Splitting data into training and testing sets
- Evaluating model accuracy, precision, recall, and F1-score
- Generating confusion matrix and classification report
## Key Findings
- Analyzed 19,384 loans, classifying them as healthy or high-risk.
- Model achieved 100% precision and F1-score for healthy loans (label 0).
- Model achieved 84% precision and 89% F1-score for high-risk loans (label 1), with 94% recall.
- Model tends to predict healthy loans extremely well but occasionally misclassifies high-risk loans as healthy.