contoso 0.1.0

Synthetic business datasets for the fictional Contoso company

contoso

Synthetic business datasets for the fictional Contoso company. This Python package provides access to 8 interconnected business datasets stored as Parquet files on Cloudflare R2, loaded efficiently via DuckDB.

Installation

pip install contoso

Or install from source:

pip install git+https://codeberg.org/usrbinr/contoso.py

Usage

import contoso

# Create a DuckDB connection with Contoso datasets
db = contoso.create_duckdb(size="small")

# Access tables (lazy evaluation via DuckDB relations)
db["sales"].limit(10).fetchdf()
db["customer"].filter("Country = 'United States'").fetchdf()

# Available tables
# - sales: Sales transaction data
# - product: Product details (name, manufacturer, category)
# - customer: Customer demographics and geography
# - store: Store locations and attributes
# - fx: Foreign exchange rates
# - calendar: Date dimension table
# - orders: Order header information
# - orderrows: Order line items

# Launch interactive web UI
contoso.launch_ui(db["con"])

# Close connection when done
db["con"].close()

Dataset Sizes

Size	Sales Rows
small	~8,000
medium	~2.3 million
large	~47 million
mega	~237 million

Requirements

• Python >= 3.9
• DuckDB >= 1.0.0

License

MIT