Predictive Credit Risk Model

This repository contains a machine learning project for credit risk prediction using the UCI Default of Credit Card Clients dataset. The model predicts whether a client will default on their credit card payment based on their demographic, payment history, and bill statement data.

---

Overview

Credit risk assessment is crucial for financial institutions to minimize losses. This project utilizes a Random Forest Classifier to predict the likelihood of a client defaulting, with results evaluated using metrics like Accuracy, ROC AUC Score, and Classification Report.

---

Dataset

The dataset used is sourced from the UCI Machine Learning Repository and contains:

• 30,000 samples of credit card clients.
• 23 features including:
  • Demographic information: SEX, AGE, EDUCATION, MARRIAGE
  • Payment history: PAY_0 to PAY_6
  • Bill statements: BILL_AMT1 to BILL_AMT6
  • Payment amounts: PAY_AMT1 to PAY_AMT6
• Target variable: default (1 = Default, 0 = No Default)

---

Workflow

1. Data Preprocessing:

   • Filling missing values with column means.
   • Standardizing numeric features using StandardScaler.
   • Encoding categorical variables using LabelEncoder.

2. Class Balance Check:

   • The dataset has an equal distribution of Default and No Default classes (4673 samples each), ensuring no need for resampling techniques.

3. Model Training:

   • A Random Forest Classifier is trained.
   • Hyperparameter tuning performed using GridSearchCV.

4. Model Evaluation:

   • Accuracy: 85.4%
   • ROC AUC Score: 0.924
   • Detailed Classification Report and Confusion Matrix are generated.

5. Feature Importance:

   • The top predictors of credit default are identified, including LIMIT_BAL, PAY_0, and BILL_AMT features.

---

Results

Key Metrics:

Metric	Value
Accuracy	85.4%
ROC AUC	0.924
Precision	0.85–0.86
Recall	0.85–0.86

Confusion Matrix:

The confusion matrix highlights the prediction performance for both classes:

Actual/Predicted	No Default	Default
No Default	4024	649
Default	711	3962

---

Installation

To run this project locally, follow these steps:

1. Clone the Repository:

   git clone https://github.com/<YourUsername>/<RepoName>.git
   cd <RepoName>

2. Install Dependencies: Install the required Python libraries using pip:

   pip install -r requirements.txt

3. Run the Jupyter Notebook: Open the Jupyter Notebook to explore the code:

   jupyter notebook

---

Requirements

• Python 3.8+
• Libraries:
  • pandas
  • numpy
  • matplotlib
  • seaborn
  • scikit-learn
  • imbalanced-learn (if SMOTE is applied in future versions)

---

Visualizations

1. Confusion Matrix:

2. Feature Importance:

---

Next Steps

• Compare performance with other models like XGBoost and LightGBM.
• Deploy the model as an API for real-time predictions.
• Add visualization dashboards for better insights.

---

Contributions

Contributions are welcome! Feel free to fork the repository, create a new branch, and submit a pull request.

---

License

This project is licensed under the MIT License.

---

Author

• Jebin Larosh Jervis
• Connect with me: LinkedIn