ChurnGuard — Live Customer Churn Prediction Platform | Open Source | Mopuru Upendra Reddy

LIVE · OPEN SOURCE · FREE · 2026

ChurnGuard

Customer Churn Prediction Platform

Real-time Analytics · ML Risk Scoring · Optional Claude AI Advisor · Python ML

// Stop Losing Customers. Start Predicting Churn.

⚛️ React 18 TypeScript 5 🐍 Python ML 🤖 Claude AI (optional) Node.js 24 🆓 Open Source

🚀 LIVE DEMO — CLICK TO OPEN

Website	churnguard-customer-churn-predictio.vercel.app
Dashboard	/dashboard →
Backend API	/api/churn/summary →
Source Code	github.com/upendra8690/churnguard-customer-churn-prediction

⚠ Backend runs on Render's free tier and sleeps when idle — first request after inactivity takes 30-50s to wake up.

940

Customers in Dataset

32.1%

Churn Rate Detected

9

API Endpoints

<1s

Prediction Speed

~68%

ML Test Accuracy

$0

Cost, Forever

📖 The Story

Most companies have no idea a customer is about to leave until the cancellation email arrives. Tools that solve this exist — but they're typically enterprise software costing tens of thousands of dollars a year, built for teams with a data scientist on staff.

I built ChurnGuard as my own version of that idea — a full-stack, open-source churn intelligence platform. A real Python scikit-learn model, a real Node.js REST API, a real React dashboard on top of it, and an optional Claude AI advisor for plain-English retention recommendations.

It's not just code sitting in a repo. It's deployed — a live site, a live API, a live dashboard — click the links above and try it yourself.

✨ Platform Features

📊

REAL-TIME DASHBOARD

Live KPI cards, 5 tabs, dark/light mode. Churn rate, revenue at risk, high-risk customers — all computed from the real dataset.

🔮

ML RISK ENGINE

Scores every customer 0–100 using contract type, tenure, charges, and demographics. A weighted model, transparent and explainable.

🤖

CLAUDE AI ADVISOR

Optional. Generates retention recommendations from live data when an Anthropic API key is set — otherwise runs in a clearly-labeled demo mode.

👥

CUSTOMER REGISTRY

940 customers. Searchable, sortable, filterable, paginated. One-click CSV export.

📈

5 CHART TYPES

Contract, internet, payment, tenure, monthly charges — visualized with Recharts.

⚡

9 REST API ENDPOINTS

Full OpenAPI 3.1 spec, no auth required. Hit the links above right now — they return real JSON.

🐍

PYTHON ML MODEL

Standalone logistic regression, scikit-learn, ~68% test accuracy. Runs from the command line too.

🌙

DARK / LIGHT MODE

System-aware theme switching across every chart and card.

🚀

ACTUALLY DEPLOYED

Frontend on Vercel, backend on Render. Not just a repo — a working link anyone can open.

🗺 Public Roadmap

✅ Q3 2026 — LIVE NOW

Foundation Complete — Everything Below is Live

No signup. No credit card. No limits. Open it and go.

✓Real-time churn analytics dashboard

✓ML weighted risk scoring (0–100)

✓Claude AI advisor (with demo-mode fallback)

✓9 REST API endpoints + OpenAPI spec

✓Python ML model (scikit-learn)

✓Deployed on Vercel + Render

🔧 Q4 2026 — COMING NEXT

Integrations & Automation

○Slack & email real-time alerts

○Salesforce & HubSpot CRM sync

○Custom risk model training UI

○PostgreSQL + Docker

Q3 2026

Live now ✅

Q4 2026

In progress 🔧

2027

Planned 🚀

Later

Ideas 🌟

💡 Why This Exists

🎯

BUILT TO BE USEFUL

Every feature exists because a churn dashboard should help someone make a real decision — not just look good in a screenshot.

🔓

FREE MEANS FREE

No trial period, no feature paywall. The full dashboard, the API, and the ML model are yours to use and modify.

🌱

BUILT SOLO, IN THE OPEN

One developer's project, shipped publicly. The roadmap above is real, and it updates as work actually gets done.

🌍

Making churn intelligence
accessible, not just enterprise-priced.

Enterprise churn tools often cost $50,000/year. ChurnGuard is free and open source — built to be capable enough for a real team to rely on, and simple enough for one person to deploy in an afternoon.

⭐ Star on GitHub 🚀 Try the Live Demo

⚡ API Reference

9 REST ENDPOINTS · churnguard-customer-churn-prediction.onrender.com/api/churn

GET /summary	Churn rate, avg charges, tenure, senior rate
GET /by-contract	Churn counts by contract type
GET /by-internet	Churn counts by internet service
GET /by-payment	Churn counts by payment method
GET /tenure-distribution	Churn across 6 tenure buckets
GET /monthly-charges	Charges distribution
GET /customers	All 940 customers with risk scores
POST /predict	Predict churn from customer attributes
POST /ai-advisor	AI retention recommendations (demo mode without a key)

🚀 Run It Yourself

1

Clone the Repository

git clone https://github.com/upendra8690/churnguard-customer-churn-prediction.git cd churnguard-customer-churn-prediction

2

Start the Backend

cd backend npm install npm run dev

3

Start the Frontend

cd frontend npm install npm run dev

4

Run the Python Model

pip install pandas scikit-learn python churn_model.py

👨‍💻 About the Builder

🧑‍💻

Mopuru Upendra Reddy

@upendra8690 · CSE Student · AI/ML · Cybersecurity · Data Science

Building full-stack, applied-ML projects to actually ship — not just prototype. ChurnGuard is deployed, open source, and free to use or fork.

⭐ GitHub 💼 LinkedIn 🐦 Twitter/X ✍️ Medium

Star It. Fork It. Use It.

Free, open source, deployed, and built to actually work — click the live links above and see for yourself.

🚀 Open Live Demo ⭐ View on GitHub

Built by Mopuru Upendra Reddy · MIT License · Open Source · 2026
github.com/upendra8690/churnguard-customer-churn-prediction