📊 ecommerce-bi-dashboard — Lazada vs Zalora Price War Analytics

Raw scraped e-commerce data is only useful if someone can actually explore it. A one-off notebook analysis answers today's question — but the moment a business user (or a recruiter reviewing the work) wants to slice the data differently, filter by category, or drill into a specific segment, it means going back to the code. There's no self-serve way to ask a new question of the data.

A multi-page BI dashboard, backed by a real PostgreSQL database, that lets anyone explore the Lazada vs Zalora price war dataset interactively — no code required:

• Live KPIs — total products, average discount, platform breakdown, at a glance
• Interactive comparison — filter by category and price range, charts update instantly
• Category drill-down — pick a category to see every product in it, sortable, with summary stats
• One-command deployment — docker compose up --build seeds the database and serves the dashboard, zero manual steps

• 📈 KPI Overview — total products, avg discount %, products on discount, per-platform breakdown (count, avg discount, avg rating)
• 🔍 Interactive Filters — category dropdown + price range slider, shared across pages, charts update live via Dash callbacks
• 📊 Price Comparison Charts — avg discount by platform, price distribution (log-scale box plot), product count by category
• 🗂️ Category Drill-Down — cross-category discount comparison, or a sortable product table for any selected category
• 🐘 PostgreSQL Backend — real relational database, not a static CSV — schema includes indexes and a discount-filtered view
• 🐳 Docker-First — Compose stack with a health-checked Postgres service and automatic, idempotent data seeding on first boot
• ♿ Accessible Color Palette — platform colors validated for colorblind-safe separation and contrast on light/dark surfaces
• ✅ Integration Tested — 14 tests running against the real database, auto-skipping cleanly if Postgres isn't reachable

git clone https://github.com/rizalcodes/ecommerce-bi-dashboard.git
cd ecommerce-bi-dashboard

docker compose up --build

That's it — Postgres starts, waits to be healthy, the app seeds the database automatically from data/raw/, and the dashboard is live at http://localhost:8050.

Port note: the database maps to host port 5433 (not 5432), since a native PostgreSQL install often already occupies 5432. This only affects host-side tools (psql, tests, local dev) — the app itself connects internally on the standard port.

python -m venv .venv
.venv\Scripts\activate        # Windows (source .venv/bin/activate on macOS/Linux)
pip install -r requirements-dev.txt
copy .env.example .env         # fill in; for the compose db use DB_PORT=5433
docker compose up -d db
python data/migrate.py         # one-time seed (idempotent, safe to re-run)
python app.py

docker compose up -d db
pytest

ecommerce-bi-dashboard/
├── app.py                  # Dash entry point, page registration
├── pages/
│   ├── overview.py          # KPI summary
│   ├── comparison.py         # Filtered charts
│   └── categories.py           # Category drill-down
├── data/
│   ├── db.py                    # SQLAlchemy engine, credentials from .env
│   ├── queries.py                 # All SQL lives here
│   ├── migrate.py                   # Idempotent seed script
│   ├── schema.sql                     # Table + index + view definitions
│   └── raw/                             # Source CSVs (Lazada + Zalora)
├── components/                            # Reusable filters, KPI cards, charts
├── tests/                                   # 14 integration tests
├── docker-compose.yml
└── Dockerfile

Rizal

Built with Plotly Dash, PostgreSQL, and the belief that a chart you can't filter is just a screenshot.