🎯 Definicja

🔑 Kluczowe punkty

📚 Szczegółowe wyjaśnienie

💡 Przykład zastosowania

 

📌 Źródła

👽 Brudnopis

Anomaly detection discovers any inconsistencies or irregularities in the data.

  • Anomaly detection is applied in the following cases:
    • After profiling the data: Anomaly detection is automatically started following each profiling run.
    • In monitoring projects, as part of the data quality (DQ) evaluation .
  • Anomaly detection relies on one of the two models:
    • Time-independent model: This model is used by default.
    • Time-dependent model, based on the time series analysis.
  • You can select time-independent or time-dependent anomaly detection to be used during profiling.
  • The model is equipped to detect unexpected values of different kinds such as unexpected nulls, negatives, positives or zeros and changes from established trends.
  • Time-dependent anomaly detection requires a minimum of 6 runs.