Churn Prediction — Interactive Dashboard

Interactive ML dashboard for customer churn analysis. Visualizes churn patterns, model predictions, feature importance, and at-risk customers with live filters.

Features

• 6 live KPI cards — churn rate, risk count, avg probability
• Churn distribution donut + risk level breakdown
• Probability histogram — retained vs churned overlap
• Feature importance from trained XGBoost model
• Churn rate by plan, region, tenure, login frequency, satisfaction
• At-risk customer table with sortable columns
• All charts filter simultaneously

Tech Stack

• Python — Dash, Plotly, Pandas, joblib
• ML — XGBoost model (AUC 0.9879)

Setup

1. Clone repo
2. python -m venv venv → activate
3. pip install -r requirements.txt
4. Place churn_data.csv in data/
5. Place model files in models/
6. python app.py
7. Open http://localhost:8051

Screenshots