Free practice · single-choice · hard
Practice
Separate profiling from cleaning
Why should the raw extract be profiled before its unexpected values are recoded?
Five-row order profile
A short profile exposes three different quality risks
- Rows
- 5 Source extract
- Distinct order IDs
- 4 One repeated key
- Approved statuses
- 3 One blank · one unexpected
Rows with a missing status20%
The duplicate key, blank status, and unapproved label must be investigated before shipped revenue is reported.