Skip to content

Latest commit

 

History

History
64 lines (54 loc) · 2.45 KB

README.md

File metadata and controls

64 lines (54 loc) · 2.45 KB
Raw

Credit Risk Assessment Model

Overview

This project involves developing a predictive model to assess loan default risks. Using a dataset that includes various borrower characteristics, the model predicts whether a loan will default based on historical data.

Table of Contents

Technologies Used

  • Python
  • Pandas
  • Scikit-learn
  • NumPy
  • Matplotlib/Seaborn (for visualizations)
  • Jupyter Notebook (for development)

Dataset

The dataset used for this project contains information about borrowers, including features like age, income, employment length, loan amount, interest rate, and loan status. It consists of 32,581 entries with various attributes relevant to credit risk.

Features

  • person_age: Age of the borrower
  • person_income: Income of the borrower
  • person_emp_length: Employment length (in years)
  • loan_amnt: Amount of the loan
  • loan_int_rate: Interest rate of the loan
  • loan_status: Status of the loan (0 for non-default, 1 for default)
  • Additional features related to borrower demographics and credit history

Modeling Approach

  1. Data Preprocessing: Clean and preprocess the data, handle missing values, and encode categorical variables.
  2. Train-Test Split: Divide the dataset into training and testing sets.
  3. Model Selection: Use logistic regression to model the likelihood of default.
  4. Hyperparameter Tuning: Optimize the model using techniques such as grid search.
  5. Evaluation: Evaluate the model's performance using accuracy, confusion matrix, and classification report.

Results

  • Accuracy: 86.49%
  • Confusion Matrix:
Predicted: No (0) Predicted: Yes (1)
Actual: No (0) True Negatives False Positives
Actual: Yes (1) False Negatives True Positives
  • Classification Report: Detailed performance metrics (precision, recall, F1-score).

Usage

  1. Clone the repository: 
    git clone https://github.com/lakshyajoshii/credit_risk_model.git
  2. Navigate to the project directory: 
    cd credit_risk_model
  3. Run the main script: 
    python main.py

License

This project is licensed under the MIT License. See the LICENSE file for details.