Inspect Data and Define Acceptance Rules
Unit ID: M10-U01 Estimated active time: 35-50 minutes
Before transforming, record file name, checksum, row meaning, columns, raw types, blanks, duplicate key, allowed categories, numeric ranges, and exclusions.
Separate parsing from acceptance. A value can parse as a number and still violate the task range.
Create a small rule table:
| Field | Required | Type | Allowed rule | Missing action |
|---|---|---|---|---|
| record ID | yes | text | unique after trimming | reject |
| numeric measure | yes | number | pathway range | reject |
Write expected row-count behaviour. Rejected rows remain visible in a separate evidence table; they are not silently dropped.