Predicting Bank Marketing Campaign Success

📝 Project Overview

This project focuses on building a predictive model to determine whether a customer will subscribe to a term deposit based on their attributes and past interactions with a bank. Using the UCI Bank Marketing Dataset, we applied machine learning techniques, explored data patterns, and derived actionable insights.

Key Highlights

• Goal: Predict term deposit subscription (yes or no).
• Dataset: UCI Bank Marketing Dataset (Link).
• Techniques: Data preprocessing, feature engineering, exploratory data analysis (EDA), hyperparameter tuning.
• Best Model: XGBoost, with an F1-Macro Score of 0.9324 and ROC-AUC of 0.9345 after hyperparameter tuning.

---

📂 Dataset Description

The dataset consists of direct marketing campaigns (phone calls) conducted by a Portuguese banking institution. The goal is to predict whether a client will subscribe to a term deposit (y).

Features

The dataset contains 17 input features:

• Client Information: age, job, marital, education, default, balance, housing, loan.
• Last Contact Details: contact, day, month, duration.
• Campaign Performance Metrics: campaign, pdays, previous, poutcome.
• Target Variable: y (yes or no for term deposit subscription).

Dataset Source

• UCI Machine Learning Repository

---

🛠️ Project Pipeline

1. Exploratory Data Analysis (EDA):

   • Visualized distributions, relationships, and correlations.
   • Analyzed class imbalance in the target variable.

2. Data Preprocessing:

   • Handled missing values and inconsistent data.
   • Encoded categorical variables using one-hot encoding and label encoding.
   • Standardized numerical features for model compatibility.

3. Feature Engineering:

   • Selected important features using tree-based importance metrics.
   • Handled class imbalance with SMOTE (Synthetic Minority Oversampling Technique).

4. Model Development:

   • Baseline model: Logistic Regression.
   • Advanced models: Random Forest and XGBoost.
   • Hyperparameter tuning using RandomizedSearchCV.

5. Evaluation:

   • Evaluated models using metrics like F1-score, ROC-AUC, and confusion matrix.
   • XGBoost yielded the best performance with optimal hyperparameters.

6. Insights and Recommendations:

   • Identified key factors influencing term deposit subscription.
   • Provided actionable recommendations for targeted marketing.

---

🔍 Key Findings

• Model Performance:

  • Best Model: XGBoost
  • F1-Macro Score: 0.9324
  • ROC-AUC Score: 0.9345

• Feature Importance:

  • The most influential features were housing (has housing loan?), marital_married (is married?), marital_single (is single?), month_mar (is March?), and month_apr (is April?).

• Class Imbalance:

  • Addressed severe class imbalance using SMOTE, ensuring the model performed well on both classes.

---

📊 Visualizations

• Confusion Matrix:
  • Showcases the performance of the best model on test data.
• Feature Importance Plot:
  • Highlights key drivers of term deposit subscription.

---

🖥️ Technology Stack

• Programming Language: Python
• Libraries:
  • Data Processing: pandas, numpy
  • Visualization: matplotlib, seaborn, missingno
  • Machine Learning: scikit-learn, XGBoost
  • Imbalance Handling: imblearn

---

📁 Repository Structure

📂 bank-marketing-prediction
├── 📄 README.md                # Project documentation
├── 📄 Analysis.ipynb           # EDA, model training, and analysis
├── 📄 best_xgboost_model.pkl   # Saved XGBoost model
├── 📊 Confusion-Matrix         # Confusion Matrix image
├── 📊 Feature-Importance       # Feature Importance image

---

🚀 How to Run

1. Clone the repository:

   git clone https://github.com/yourusername/bank-marketing-prediction.git
   cd bank-marketing-prediction

2. Install dependencies:

   pip install -r requirements.txt

3. Run the Jupyter Notebooks for EDA and Model Training:

   jupyter notebook eda_notebook.ipynb
   jupyter notebook model_training.ipynb

---

🔗 References

• UCI Bank Marketing Dataset
• XGBoost Documentation
• Scikit-learn Documentation

---

✍️ Authors

• Aditya Kulkarni

  • LinkedIn: @aditya-s-kulkarni
  • GitHub: @Aditya-k-23
  • Website: adityakulkarni.me

• Anaqi Amir

  • LinkedIn: @anaqi-amir
  • GitHub: @AnaqiAmir