LOCAL_POSTGRES.md

Local PostgreSQL workflow

This project uses Docker Compose to run PostgreSQL on your own computer. It is a safe way to practise a real database flow without paying for a cloud provider or exposing credentials.

sales.csv → Pandas cleaning + Pandera validation → PostgreSQL → Pandas → dashboard

Start and load data

From the project root:

docker compose up -d
python src/run_pipeline.py --load-postgres

The first command starts a PostgreSQL container on port 5433; the second one cleans the CSV, writes the data-quality report and replaces the three analytics tables in PostgreSQL.

Run the dashboard against PostgreSQL

DATABASE_URL="postgresql+psycopg://analytics_user:analytics_dev_password@localhost:5433/analytics" \
streamlit run dashboard/app.py

The sidebar will show Source: PostgreSQL. Without DATABASE_URL, the app automatically reads its committed CSV fallback instead.

Refresh after changing source data

python src/run_pipeline.py --regenerate --load-postgres

This performs the complete local refresh in one command: regenerate source data, clean it, validate it, update reports and load PostgreSQL.

Stop or reset

docker compose down

Stopping the container does not remove its stored data. To deliberately remove the local database volume as well, use docker compose down -v; this deletes only this project's local PostgreSQL data.