Predictive Modeling of Bank Term Deposits — A Kaggle Competition Project

---

📌 Overview

This project is part of the Kaggle Playground Series S5E8 (2025)

The challenge is to predict whether a client will subscribe to a bank term deposit based on tabular features.

The competition provides an approachable, real-world inspired dataset for practicing supervised machine learning workflows.

🎯 Goal

• Objective: Predict the probability that a client subscribes to a term deposit (y).

• Problem Type: Binary Classification

• Evaluation Metric: Area Under the Receiver Operating Characteristic Curve (ROC-AUC)

✨ Key Highlights

• Conducted comprehensive Exploratory Data Analysis (EDA) with visualizations.

• Implemented data preprocessing: missing values imputation, categorical encoding, and feature scaling.

• Developed and compared multiple machine learning models:

  • Logistic Regression (baseline model)

  • LightGBM Classifier

  • XGBoost Classifier

• Evaluated models using ROC-AUC, confusion matrices, and classification reports.

• Produced a submission file with probability predictions for Kaggle evaluation.

🛠️ Tools & Libraries

• Language: Python

• Core Libraries:

  • numpy, pandas → Data manipulation

  • matplotlib, seaborn → Data visualization

  • scikit-learn → Preprocessing, Logistic Regression, model evaluation

  • lightgbm → LightGBM Classifier

  • xgboost → XGBoost Classifier

📂 Dataset

The dataset is synthetically generated from the original Bank Marketing Dataset.

• train.csv → Training dataset (with target y)

• test.csv → Test dataset (no labels, predictions required)

• sample_submission.csv → Correct submission format

Target variable:

• y = 1 → Client subscribes

• y = 0 → Client does not subscribe

📊 Results

Validation ROC-AUC performance across models:

Rank	Model	ROC-AUC Score
1	LightGBM	0.9680
2	XGBoost	0.9672
3	Logistic Regression	0.9346

• LightGBM achieved the highest validation performance, closely followed by XGBoost.

• Logistic Regression, while less accurate, served as a strong and interpretable baseline.

📁 Repository Structure

├── notebooks/
│   ├── Binary_Classification_Bank_Dataset.ipynb   # Full project workflow
├── data/
│   ├── train.csv
│   ├── test.csv
│   ├── sample_submission.csv
├── results/
│   ├── submission.csv
├── requirements.txt
└── README.md

📚 Citation

Walter Reade and Elizabeth Park.
Binary Classification with a Bank Dataset.
Kaggle Playground Series S5E8 (2025)

---

🚀 Next Steps

• Perform hyperparameter tuning with GridSearchCV / Optuna
• Explore advanced feature engineering for categorical variables
• Compare additional models (CatBoost, Random Forest, Neural Networks)
• Deploy the best-performing model using Streamlit or Flask
• Build a Power BI / Tableau dashboard for interactive insights

---

📄 License

This project is licensed under the MIT License - you are free to use, modify, and distribute this project with proper attribution.