Outlier and Anomaly Investigation Guide
Separate errors from real extremes.
The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.
ROLE: You are a data quality analyst. CONTEXT: Column: [COLUMN_NAME] in [DATASET_DESCRIPTION]. Summary statistics: [SUMMARY_STATS]. The top 20 highest and lowest values with their row context: [PASTE_EXTREME_ROWS] Domain limits I know: [DOMAIN_LIMITS]. STEPS: 1. Classify each extreme as likely data entry error, unit mismatch, duplicate, legitimate rare event, or unknown, with evidence from the row context. 2. Recommend a detection rule (interquartile range, z-score, or domain limits) and justify it for this distribution; show the spreadsheet formula. 3. Propose treatment per class (correct, exclude, cap, keep and flag), who to ask for confirmation, and how to document decisions. 4. List questions to send to the data owner for the unknown cases. OUTPUT FORMAT: Classification table, detection rule with formula, treatment plan, and owner questions. QUALITY BAR: Do not delete outliers by default; real extremes can be the most important rows. Document every decision. FOLLOW-UP: Ask me what an impossible value looks like in this column.
Fill in before use
Optional: type your details and the prompt above updates. Nothing is saved or sent.
Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.