Skip to main content

Generate your anomaly test with Elementary AI

Let our Slack chatbot create the anomaly test you need.
 elementary.volume_anomalies Monitors the row count of your table over time per time bucket (if configured without timestamp_column, will count table total rows). Upon running the test, your data is split into time buckets (daily by default, configurable with the time bucket field), and then we compute the row count per bucket for the last training_period days (by default 14). The test then compares the row count of each bucket within the detection period (last 2 days by default, configured as detection_period), and compares it to the row count of the previous time buckets. The test will only run on completed time buckets, so if you run it with daily buckets in the middle of today, the test would only count yesterday as a complete bucket. If there were any anomalies during the detection period, the test will fail.

​
Test configuration

No mandatory configuration, however it is highly recommended to configure a timestamp_column. 
data_tests:
  - elementary.volume_anomalies:
    arguments:
      timestamp_column: column name
      where_expression: sql expression
      anomaly_sensitivity: int
      anomaly_direction: [both | spike | drop]
      detection_period:
        period: [hour | day | week | month]
        count: int
      training_period:
        period: [hour | day | week | month]
        count: int
      time_bucket:
        period: [hour | day | week | month]
        count: int
      seasonality: day_of_week
      fail_on_zero: [true | false]
      ignore_small_changes:
        spike_failure_percent_threshold: int
        drop_failure_percent_threshold: int
      detection_delay:
        period: [hour | day | week | month]
        count: int
      anomaly_exclude_metrics: [SQL expression]
models:
  - name: < model name >
    data_tests:
      - elementary.volume_anomalies:
          arguments:
            timestamp_column: < timestamp column >
            where_expression: < sql expression >
            time_bucket: # Daily by default
              period: < time period >
              count: < number of periods >