📊 Credit Mix Classification – End-to-End Machine Learning Project

📌 Project Overview

Credit Mix Classification is an end-to-end machine learning project that predicts a customer’s Credit Mix category — Good, Standard, or Bad — using structured financial and behavioral data. The project covers the complete ML lifecycle, including data cleaning, exploratory data analysis (EDA), feature engineering, outlier handling, model-specific preprocessing pipelines, and comparative evaluation of multiple ensemble models.

🗂️ Project Structure

├── CSV_Files/
│ ├── Raw and cleaned datasets used for modeling
│
├── images/
│ ├── EDA plots, barplots, confusion matrices and so on.
│
├── EDA_report.html
│ ├── Automated profiling report (ydata-profiling)
│
├── Predicting Customer Credit Mix - End-to-End.ipynb
│ ├── Main notebook containing the full ML workflow
│
├── README.md
│ ├── Project documentation

📁 Cleaned Dataset Description

• Total records: 50,000
• Total features: 30
• Target variable: Credit_Mix

Target variable --> Classes

• Good
• Standard
• Bad

Feature Types

• Numerical: 'Age', 'Annual_Income', 'Monthly_Inhand_Salary', 'Num_Bank_Accounts', 'Num_Credit_Card', 'Interest_Rate', 'Num_of_Loan', 'Debt_Consolidation_Loan', 'Home_Equity_Loan', 'Student_Loan', 'Payday_Loan', 'Personal_Loan', 'Auto_Loan', 'Mortgage_Loan', 'Credit-Builder_Loan', 'Num_of_Loan_Types', 'Delay_from_due_date', 'Num_of_Delayed_Payment', 'Changed_Credit_Limit', 'Num_Credit_Inquiries', 'Outstanding_Debt', 'Credit_Utilization_Ratio', 'Credit_History_Months', 'Total_EMI_per_month', 'Amount_invested_monthly', 'Monthly_Balance'.
• Categorical: 'Occupation', 'Payment_of_Min_Amount', 'Payment_Behaviour_lavel', 'Payment_Behaviour_size'.

🔍 Exploratory Data Analysis (EDA)

EDA was performed using both manual visualizations and automated profiling. 📄 EDA Report: EDA_report.html

Key Analysis Steps

• Distribution analysis
• IQR-based outlier analysis
• Feature correlation analysis
• Class imbalance inspection

🧹 Data Preprocessing & Feature Engineering

• Missing value handling
• Outlier detection and treatment
• Robust scaling for extreme numerical outliers
• Ordinal encoding for categorical features
• Model-specific preprocessing pipelines
• Class imbalance handling using class weights

🤖 Machine Learning Models

The following ensemble models were trained and evaluated:

• Random Forest
• XGBoost
• LightGBM

📈 Model Evaluation

Models were evaluated using multiple performance metrics:

• Training Accuracy
• Test Accuracy
• Precision (Weighted)
• Recall (Weighted)
• F1-Score (Weighted)
• Confusion Matrix

📊 Performance Summary (Test Set)

Model	Accuracy	F1-Score	ROC-AUC
Random Forest	0.966933	0.966884	0.997442
XGBoost	0.970933	0.970923	0.998082
LightGBM	0.964933	0.964884	0.997703

🧠 Key Insights

• Gradient boosting models slightly outperformed Random Forest.
• RobustScaler significantly improved performance on income-related features.
• Class-weight handling effectively addressed class imbalance.

🛠️ Tools & Technologies

• Programming Language: Python
• Data Processing: Pandas, NumPy
• Machine Learning: Scikit-learn, XGBoost, LightGBM,
• Visualization: Matplotlib, Seaborn
• EDA: ydata-profiling

🚀 How to Run the Project

1. Clone the repository
2. Install required dependencies
3. Open the notebook
4. Run cells sequentially

👤 ---- Author ----

Shahriar Hussain
Machine Learning & Data Science Practitioner