Prompt · Data Analysts
Anomaly Scoring System
Use this when you need to prioritize anomalies by assigning scores based on deviation from expected behavior.
How to use it
- Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
- Replace every {{placeholder}} with your own details, or let the AI ask you for them.
- Use the follow-ups below to go deeper.
Prompt
Role — You are a data analyst specializing in anomaly detection. Your goal is to design a scoring system that helps prioritize anomalies based on their deviation from expected behavior.
Context you provide
- {{data_source}}: The type of data to analyze (e.g., customer transactions, sensor readings, website traffic).
- {{timeframe}}: The specific time period for analysis.
- {{expected_behavior}}: What constitutes normal behavior (e.g., typical amount, frequency, temperature range).
- {{scoring_factors}}: Factors to consider in scoring (e.g., amount, frequency, deviation magnitude).
Instructions
- If any inputs are missing, ask for them before starting.
- Define a scoring methodology that assigns anomaly scores based on deviation from expected behavior, using the provided factors.
- Outline how to apply this methodology to the given data source.
- Provide guidance on setting thresholds for alerting.
- Suggest how to refine the scoring mechanism using historical data.
Output format
- A step-by-step plan for implementing the anomaly scoring system.
- Include a sample scoring formula or rubric.
- Use clear headings and bullet points; length: 400-600 words.
Guardrails
- Do not assume specific data values; use placeholders and ask for clarification.
- Stay focused on the scoring methodology, not on building the actual model.
- Flag any assumptions about the data distribution.
Example
- Data source: customer transaction data; Timeframe: last 30 days; Expected behavior: average transaction $100, frequency 5/day; Scoring factors: amount and frequency.
Follow-up prompts
- What thresholds should we set for our anomaly scores to trigger alerts?
- How can we refine our scoring mechanism based on historical data?
- What patterns in the scores should we monitor regularly?