Everything useful before you download or buy.
Ten checks to run on any new table before you trust a number from it, a minute each.
A printable two-page A4 sheet. Each check has the SQL (DuckDB) and a pandas one-liner, what a bad result looks like, and a worked run on the agricultural exports and prices from Can I Trust This Data?, with the real counts: rows against what was promised, column types, exact and key duplicates, missing values, spellings, ranges, values far from their group's median, gaps and freshness, and the master list.
On the book's data the checks find, among other things, 3 exact and 33 key duplicates, seven spellings of four countries, a tonnage of −4,500, two prices about ten times too high and 63 export rows with no price in their year.
It ends with a tick-box summary to fill in for your own table and a five-line template for the note you write afterwards. PDF, two A4 pages.
Read the complete resource here once you are signed in, or download it below.
One review per person. Submitting again updates your existing review.
Be the first to review this resource.