Skip to content

mharis8598/telco-churn-predictor

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Key Findings

EDA Insights

  • Month-to-month contracts have a 42.7% churn rate vs just 2.8% for two-year contracts
  • Fiber optic users churn at 41.9% — more than double DSL users (19%)
  • Electronic check payers churn at 45.3%, far above other payment methods
  • Customers without tech support or online security churn above 40%
  • New customers (tenure < 5 months) are the most likely to leave

Model Performance

Model ROC AUC Accuracy Churn Recall Churn Precision
Logistic Regression 0.821 0.74 0.71 0.51
Random Forest 0.816 0.76 0.63 0.54
XGBoost (Baseline) 0.805 0.76 0.65 0.54
XGBoost (Tuned) 0.811 0.77 0.64 0.55

Top Churn Drivers (Feature Importance)

  1. Two-year contract (strongest retention factor)
  2. Fiber optic internet service
  3. Electronic check payment method
  4. One-year contract
  5. Tenure (longer = less likely to churn)

Screenshots

Churn Predictor App ROC Comparison Feature Importance

About

End-to-end ML pipeline predicting telecom customer churn with XGBoost, featuring EDA, model comparison, and an interactive Streamlit app

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors