Data cleaning is not cosmetic preparation. Duplicate records, inconsistent codes, invalid dates, impossible values and missing observations can alter analytical results, so cleaning requires explicit rules and analytical judgement.
Profile before editing
Review variable names and types, ranges, frequencies, missingness, duplicates and unexpected categories. For numerical fields, inspect distributions and extreme values rather than assuming every unusual observation is an error.
Apply explicit cleaning rules
Standardise dates, labels and categories using documented rules. Investigate duplicates before removal, and return to source records where possible when values appear impossible rather than guessing what was intended.
Handle missing information deliberately
The appropriate treatment depends on why data are missing, how much is missing and the analysis that follows. Cleaning should not silently replace uncertainty with invented information.
Preserve an audit trail
Keep the original dataset unchanged and create a reproducible cleaned version. Record transformations, exclusions and derived variables. Formulas, scripts or documented procedures are preferable to unexplained manual edits.
Validate after cleaning
Repeat key checks after transformations. Confirm categories remain valid, row counts are understood, formulas have not introduced errors and important distributions remain plausible. Reconcile against trusted totals where appropriate.
Interpret only what the evidence supports
Cleaning improves consistency but cannot repair weaknesses in the original design. Interpretation must still consider coverage, measurement, sampling and missingness.
Build practical capability
Study the full workflow in the Data Analytics course and the DatalytIQs data-cleaning pathway.