Credit Score Classification

A machine learning pipeline that classifies a customer's credit score category from financial and behavioral data.

Overview

• Dataset: 100,000 records, 22 features (income, credit history, payment behavior, etc.)
• Data cleaning: de-duplicated and cleaned down to 86,932 high-quality records used for modeling
• Task: multi-class classification of credit score bracket

Workflow

1. Exploratory Data Analysis (EDA) — null values, duplicates, distributions
2. Feature engineering to surface key credit-risk indicators
3. Train/test split
4. Model comparison: Logistic Regression, Decision Tree, Random Forest
5. Evaluation on held-out test data to guard against overfitting

Results

Model	Accuracy
Logistic Regression	53%
Decision Tree	70%
Random Forest (best)	77.4%

Tech Stack

Python · Scikit-learn · Pandas · NumPy · Seaborn · Matplotlib

Setup

pip install -r requirements.txt

Place credit_score.csv in the project root, then run the notebook.

Possible Extensions

• Hyperparameter tuning via GridSearchCV / Optuna
• Try gradient boosting (LightGBM/XGBoost) for comparison
• SHAP-based feature importance for model explainability