E-Commerce Customer Behaviour EDA Platform

A production-style analytics platform for investigating e-commerce browsing, conversion, revenue, customer behaviour, data quality, and purchase propensity. It combines a polished React dashboard with a Flask analytics API backed by SQLite.

Included capabilities

• Executive dashboard with revenue, customer, order, conversion, bounce, session, cart, purchase and rating metrics
• Responsive analysis workspace with dark/light themes, command palette (Ctrl/Cmd + K), notifications, animated cards, and mobile navigation
• CSV upload, searchable dataset preview, pagination, schema and quality statistics
• EDA descriptive statistics, distribution charts, correlation heatmap, purchase funnel, and insight engine
• K-Means segments, PCA projection, Random Forest prediction metrics, feature importance, and confusion matrix
• Filter-aware CSV, Excel and executive PDF exports
• A deterministic 1,500-row sample dataset automatically seeded into SQLite on first API request

Architecture

frontend/     React + Vite + TypeScript UI and Recharts visualizations
backend/      Flask APIs, SQLite schema, seeded data, EDA and ML routines
backend/data/ Reserved for externally sourced CSV data snapshots

Run locally

Prerequisites: Node.js 20+ and Python 3.10+.

npm run install:all
npm run server

In a second terminal:

npm run dev

Open http://localhost:5173. The frontend proxies /api calls to Flask on port 5000.

Dataset contract

The seeded data includes customer ID, age, gender, location, device, session duration, pages viewed, category, cart value, purchase amount/status, payment method, coupon flag, returning flag, bounce, date, discount, rating, traffic source, and review.

For upload, CSV headers should include at least customer_id, purchase_amount, and purchase_status; retaining the full seeded schema enables every analytic view.

Deployment

• Build the frontend with npm run build; deploy frontend/dist to any static host.
• Run Flask behind Gunicorn or Waitress, setting the static host's API proxy target to Flask.
• Mount a persistent volume for backend/commerce.db when deployed in a container or PaaS.

pip install gunicorn
gunicorn --bind 0.0.0.0:5000 backend.app:app

License

MIT. Built as an analytics portfolio and deployment-ready starter.