Bank Marketing Campaign Analysis

Overview

This project analyzes a Portuguese banking institution's marketing campaigns to predict whether a client will subscribe to a term deposit. The analysis compares the performance of various machine learning classifiers (K-Nearest Neighbors, Logistic Regression, Decision Trees, and Support Vector Machines) to identify the most effective model for predicting customer subscription behavior.

You can view the detailed analysis in the Jupyter Notebook.

Data Source

The dataset comes from the UCI Machine Learning repository and contains information about:

• Client demographics (age, job, marital status, education)
• Campaign information (contact type, month, day, duration)
• Previous campaign outcomes
• Economic indicators

Key Findings

Model Performance

• Decision Tree emerged as the best performer with ROC-AUC score of 0.759
• KNN showed high accuracy (89.5%) but lower ROC-AUC
• Logistic Regression demonstrated stable but lower overall performance

Important Features

The most influential features for predicting term deposit subscription were:

1. Previous campaign outcome
2. Age and age groups
3. Marital status
4. Contact type
5. Month of contact

Campaign Insights

• Success rates vary significantly by month
• Previous contact outcomes strongly influence success probability
• Age and marital status are key demographic factors

Recommendations

Campaign Optimization

1. Timing:

   • Focus campaigns during months with historically higher success rates
   • Optimize contact timing based on day-of-week patterns

2. Customer Targeting:

   • Prioritize customers with positive previous interactions
   • Focus on age groups showing higher conversion rates
   • Tailor approach based on marital status and job type

3. Process Improvements:

   • Implement real-time prediction system using Decision Tree model
   • Monitor and update model periodically
   • Consider collecting additional relevant features

Repository Structure

• bank_campaigns.ipynb: Main Jupyter notebook containing analysis
• banking.py: Python script with model implementation
• images/: Directory containing visualization plots
  • Feature importance plot
  • Decision tree visualization
  • ROC curves comparison
  • Distribution plots

Requirements

• Python 3.x
• Required packages:
  • pandas
  • numpy
  • scikit-learn
  • matplotlib
  • seaborn

Usage

1. Clone the repository
2. Install required packages: pip install -r requirements.txt
3. Run the Jupyter notebook: jupyter notebook bank_campaigns.ipynb

Future Enhancements

1. Technical Implementation:

   • Deploy model as API for real-time predictions
   • Create dashboard for campaign performance monitoring
   • Set up automated model retraining pipeline

2. Business Integration:

   • Train marketing team on model insights
   • Develop targeted scripts based on key features
   • Set up A/B testing framework

Author

Tyler Scharf https://www.linkedin.com/in/tylerscharf

License

This project is licensed under the MIT License - see the LICENSE file for details.