Exacting Standards: Considerations for Precision and Accuracy ...
summary
Covers concrete rules and failures: prefer Postgres numeric over double precision for exact math, avoid the built-in money type for currency, and use PostGIS for geodata instead of freeform text. Shows real bugs — a plus sign turned a $100 gift card into an $800k liability — and explains common data errors: deletions, transcriptions, duplicated records, UTF‑8 char mismatches, timezone traps and creative nulls. Recommends types that match intent and operational checks so stored values and downstream analysis stay accurate and auditable.
AI-written summary · about
chapters (17)
- 0:00Introduction and welcome
- 1:53What is good data
- 5:10Data pipeline and types
- 6:17Storing numbers accurately
- 9:57Math pitfalls with numbers
- 12:04Text data quality issues
- 16:22Duplicate data dangers
- 21:43Time and accuracy challenges
- 24:30Boolean data best practices
- 26:17Null values and their traps
- 30:31Q&A on numeric types
- 33:01Text field constraints debate
- 35:51Boolean alternatives and trade-offs
- 37:37Horror stories and data duplication
- 40:53Merging conflicting data
- 42:43Character sets and encoding
- 44:29Quantifying data accuracy
related talks
19mLet me show you why data types still matter
46mDon't do that!
52mHow to Accurately Store Money in a Database- 19mLet me show you why data types still matter
31mCode like an accountant: Designing data systems for accuracy, resilience and auditability
19mNO ARRAY OF SUNSHINE: an introduction to data types from a data quality perspective
30mLesser known features of PostgreSQL
45mAdvanced Data Types in PostgreSQL