Comprehensive NBA database: 131 endpoints from nba_api normalized into a star schema with 13 dimensions, 20 facts, 2 bridges, 15 derived aggregations, and 4 analytics views (~58 tables).
| Format | Description |
|---|---|
| SQLite | Portable single-file database |
| DuckDB | Columnar analytics engine |
| Parquet | Compressed columnar files (zstd, partitioned by season) |
| CSV | Universal flat files |
pip install nbadb
# Full build from scratch
nbadb init
# Daily incremental update
nbadb daily
# Export to all formats
nbadb export
# Upload to Kaggle
nbadb upload| Command | Positional args | Options |
|---|---|---|
nbadb init |
— | --data-dir/-d, --format/-f (repeatable), --season-start/-s (default: 1946), --verbose/-v, --quality-check |
nbadb daily |
— | --data-dir/-d, --verbose/-v, --quality-check |
nbadb monthly |
— | --data-dir/-d, --verbose/-v, --quality-check |
nbadb full |
— | --data-dir/-d, --verbose/-v, --quality-check |
nbadb migrate |
— | --data-dir/-d |
nbadb run-quality |
— | --data-dir/-d, --report-path |
nbadb export |
— | --data-dir/-d, --format/-f (repeatable) |
nbadb upload |
— | --data-dir/-d, --message/-m (default: "Automated update") |
nbadb download |
— | --data-dir/-d |
nbadb extract-completeness |
— | --output-dir/-o, --require-full (exit non-zero when non-covered endpoints remain) |
nbadb docs-autogen |
— | --docs-root (default: docs/content/docs) |
nbadb schema [TABLE] |
optional TABLE |
— |
nbadb status |
— | --data-dir/-d, --output-format/-f (default: text; json supported) |
nbadb ask QUESTION |
required QUESTION |
--limit/-l (default: 10), --verbose/-v |
| Component | Technology |
|---|---|
| DataFrames | Polars 1.38 |
| Validation | Pandera (Polars) |
| SQLite ORM | SQLModel |
| Analytics DB | DuckDB 1.4 |
| HTTP/Proxy | proxywhirl |
| CLI | Typer + Rich |
| Docs | nbadb.w4w.dev |
Full documentation at nbadb.w4w.dev including:
- Star schema reference and ER diagrams
- Field-level data dictionary
- Column-level lineage
- DuckDB query cookbook
- Parquet usage guides
MIT
