Messy records.
WHERE IT HELPS
Standardising inconsistent entries Splitting combined fields Reformatting dates and numbers Identifying duplicates Categorising entries
FOR STANDARDISING
"Here are inconsistent entries for the same thing. Produce a standardised version, and list what you changed."
That second part matters. You need to see what was altered.
FOR DUPLICATES
"Identify likely duplicates in this list and explain why each pair matches."
Check each. Similar is not the same.
FOR CATEGORISING
"Assign each of these to one of these categories: [list]. Flag anything that does not fit."
Check the flagged ones and a sample of the rest.
WHAT TO ALWAYS DO
Keep the original. Work on a copy.
Data cleaning that goes wrong is difficult to undo.
WHAT TO NEVER DO
Accept a cleaned dataset without spot-checking Let it infer missing values
FOR MISSING DATA
"Do not fill gaps. Mark them as missing."
