Project: Credit Risk Analysis

Credit risk analysis and default prediction based on borrower and loan information. The complete workflow (EDA, data preparation, feature engineering, and model training/evaluation) is documented in a notebook.

Repository Contents

• credit_risk.ipynb: main project notebook (explanations + analysis + models).
• credit_risk_dataset.csv: dataset in CSV format.
• function_definitions.py: auxiliary functions used in the notebook (metrics and dummies).
• requirements.txt: Python dependencies.

Dataset

• Source: Kaggle, Credit Risk Dataset (https://www.kaggle.com/datasets/laotse/credit-risk-dataset)
• Target: loan_status (0/1)

Requirements

• Python 3.10+ (recommended)

Installation

Create a virtual environment and install dependencies:

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Execution

Open and run the notebook:

jupyter notebook credit_risk.ipynb

Alternatively (if using JupyterLab):

jupyter lab

What the Notebook Includes

• Dataset overview
• Data wrangling
• Exploratory data analysis (univariate/bivariate/multivariate)
• Principal Component Analysis (PCA)
• Feature engineering and final preparation
• Classification models (KNN, Decision Trees, Random Forest, Logistic Regression, SVM)
• Model comparison and improvement with boosting

Tests (optional)

If you add/have tests with pytest, you can run:

pytest -q