How venkatkudipudi009/customer-churn-prediction-using-explainable-ai works
This repository is an offline telecom churn-analysis and retention-planning project, not a deployed prediction service. The documented workflow runs in the research notebook: it processes the IBM dataset, trains and compares models, explains predictions, and produces high-risk customer outputs. The saved model artifacts and Power BI dashboard are also important deliverables. The README documents these capabilities, but the notebook internals and dashboard data connections were not sampled, so those links are left unspecified.
Main entry points: customer-churn-prediction-research-ipynb.ipynb Research notebook · IBM_telco_customer_churn.csv Telco churn data · high_risk_customers.csv High-risk customers · Models Saved models
Components 14
Analysis workflow
- Research notebook
Notebook/customer-churn-prediction-research-ipynb.ipynb - Telco churn data
Dataset/IBM_telco_customer_churn.csv - Clean and explore
Notebook/customer-churn-prediction-research-ipynb.ipynb - Feature preparation
Notebook/customer-churn-prediction-research-ipynb.ipynb - Model training
Notebook/customer-churn-prediction-research-ipynb.ipynb - Ensemble evaluation
Notebook/customer-churn-prediction-research-ipynb.ipynb - SHAP explanations
Notebook/customer-churn-prediction-research-ipynb.ipynb
Model and outputs
- Saved models
Models - Churn predictions
Outputs/final_predictions.csv - High-risk customers
Outputs/high_risk_customers.csv
Business consumption
- Power BI dashboard
Dashboard/Customer_Churn_Dashboard.pbix
Other components
- Data analyst
- Telco customers
- Retention team
Connections 12
- Data analyst → Research notebook: runs
- Research notebook → Telco churn data: reads
- Research notebook → Clean and explore: performs
- Clean and explore → Feature preparation: prepares
- Feature preparation → Model training: feeds
- Model training → Ensemble evaluation: compares
- Ensemble evaluation → SHAP explanations: explains
- Research notebook → Saved models: saves
- Research notebook → Churn predictions: writes
- Research notebook → High-risk customers: identifies
- Retention team → High-risk customers: uses
- Telco customers → Telco churn data: represented in