Data Anomalies – Automated Detection¶

Purpose¶

Catch anomalies without writing rules.

Technical Features¶

Metrics analyzed¶

• Record volume
• Missing values
• Distributions & histograms
• Value ranges
• Uniqueness

Intelligent detection¶

• Uses historical learning to define expected ranges dynamically
• Flags anomalies when actual data falls outside expected boundaries

Detection Scenarios¶

• Volume drops/spikes → e.g., missing half of daily transactions
• Column swaps → first name and last name columns reversed
• Unexpected values → “Zurich” showing up in Austrian cities

Value¶

Automates what would normally require hundreds of manual rules.