Customer Churn

Predicting customer attrition in the telecommunications industry using ensemble learning. This project focuses on identifying the key drivers of churn to improve business retention strategies.

🔗 Original Notebook on Kaggle: View here

🛠️ Technical Workflow

• Exploratory Data Analysis (EDA): Statistical analysis using Chi-square tests and correlation matrices to select features.
• Data Cleaning: Smart removal of redundant features (like gender and TotalCharges) to simplify the model.
• Feature Engineering: - Manual Ordinal Encoding for contract types (Month-to-month, One year, Two year).
  • One-Hot Encoding for categorical services.
  • Standard Scaling for numerical variables (tenure, MonthlyCharges).
• Modeling: Implementation of a Random Forest Classifier with class_weight='balanced' to handle dataset imbalance without oversampling.

📈 Key Insights

• Contract Type: Monthly contracts are the highest indicator of potential churn.
• Tenure: Customer loyalty (length of stay) inversely correlates with the probability of leaving.

🚀 How to use

1. Clone the repo: git clone https://github.com/lucalullo/customer-churn-prediction.git
2. Install dependencies: pip install pandas numpy seaborn matplotlib scikit-learn
3. Run the customer-churn.ipynb notebook.

---

Author: Luca Lullo Data Scientist | Machine Learning Applied