Skip to main content

SQL and analytics

Inspect unfamiliar files and answer questions without writing a program.

Understand a newly received dataset​

undatum table sales.csv --limit 20
undatum tui sales.csv
undatum web sales.csv
undatum stats sales.csv
undatum frequency sales.csv --fields region,status

tui and web need extras: pip install "undatum[tui]" / "undatum[web]".

Filter without writing a query​

--where takes a SQL condition on select, search, head, sample, count and convert and keeps the original values in the output:

undatum count sales.csv --where "amount > 100"
undatum select sales.csv --where "amount > 100" --add "net = amount * 0.8" --fields name,amount,net

Ad-hoc SQL across files​

A single input is the view data; multiple inputs are named after their file stems.

undatum sql "SELECT region, SUM(amount) AS total FROM data GROUP BY 1" sales.parquet \
  --output totals.csv --format-out csv

undatum sql "SELECT * FROM orders JOIN users USING (user_id)" orders.csv users.parquet

Plot​

pip install "undatum[plot]"
undatum plot sales.csv --field amount --type histogram --output amount.png

See sql, stats, plot, tui, and web.