Keep raw practice data and an exclusion log
An exclusion needs a checked reason while the original data remains available.
Separate the unchanged dataset from its working copy and record each exclusion. Equal numbers alone do not prove a duplicate; they might be independent repeated observations. The reason must identify the particular record, not merely an inconvenient result.
What to do
- Keep the original dataset separately and prepare a working copy with stable row identifiers.
- Verify the reason from the exercise instructions or a confirmed record, rather than removing matches automatically.
- Exclude only the confirmed row from the copy and log its identifier, reason and remaining membership.
Example
Fictional Raw-A contains A, B, C and D. The exercise explicitly confirms D was copied from B by mistake, not observed independently. Raw-A retains 4 rows; Clean-A excludes 1 and keeps A, B, C: 3 rows.
My Day · Windows
Save the log in My Day; data files and cleaning remain your separate work.