Column Anomalies - Paradime

The elementary.column_anomalies test executes column-level monitors and anomaly detection on a specific column. It checks the data type of the column and only executes monitors that are relevant to it.

How it works

The test analyzes the specified column in the table. It can analyze as many columns as you specify.
Based on the data type of the column, it applies relevant monitors.
You can specify which monitors to run using the column_anomalies parameter.

Default Monitors by Data Type

Data quality metric	Column Type
`null_count`	any
`null_percent`	any
`min_length`	string
`max_length`	string
`average_length`	string
`missing_count`	string
`missing_percent`	string
`min`	numeric
`max`	numeric
`average`	numeric
`zero_count`	numeric
`zero_percent`	numeric
`standard_deviation`	numeric
`variance`	numeric

Opt-in monitors by type:

Data quality metric	Column Type
`sum`	numeric

Models
Models example

models:
  - name: < model name >
    config:
      elementary:
        timestamp_column: < timestamp column >
    columns:
      - name: < column name >
        tests:
          - elementary.column_anomalies:
              column_anomalies: < specific monitors, all if null >
              where_expression: < sql expression >
              time_bucket: # Daily by default
                period: < time period >
                count: < number of periods >

  - name: < model name >
    ## if no timestamp is configured, elementary will monitor without time filtering
    columns:
      - name: < column name >
        tests:
          - elementary.column_anomalies:
              column_anomalies: < specific monitors, all if null >
              where_expression: < sql expression >

models:
  - name: login_events
    config:
      elementary:
        timestamp_column: 'loaded_at'
    columns:
      - name: user_name
        tests:
          - elementary.column_anomalies:
              column_anomalies:
                - missing_count
                - min_length
              where_expression: "event_type in ('event_1', 'event_2') and country_name != 'unwanted country'"
              time_bucket:
                period: day
                count: 1
              tags: ['elementary']

  - name: users
    ## if no timestamp is configured, elementary will monitor without time filtering
    tests:
        elementary.volume_anomalies
          tags: ['elementary']
    columns:
      - name: user_id
        tests:
          - elementary.column_anomalies:
              tags: ['elementary']
              timestamp_column: 'updated_at'
              where_expression: "event_type in ('event_1', 'event_2') and country_name != 'unwanted country'"
              time_bucket:
                period: < time period >
                count: < number of periods >
      - name: user_name
        tests:
          - elementary.column_anomalies:
              column_anomalies:
                - missing_count
                - min_length
              tags: ['elementary']

Test configuration

tests:
  — elementary.column_anomalies:
    column_anomalies: column monitors list
    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
    detection_delay:
      period: [hour | day | week | month]
      count: int
    ignore_small_changes:
      spike_failure_percent_threshold: int
      drop_failure_percent_threshold: int
    anomaly_exclude_metrics: [SQL expression]

Important Notes

No mandatory configuration, however, it is highly recommended to configure a timestamp_column.
Use column_anomalies to specify which monitors to run (if not specified, all default monitors will run).
The where_expression can be used to filter the data being tested.
If no timestamp is configured, Elementary will monitor without time filtering.
Tags can be used to run elementary tests on a dedicated run.
You can configure the test at the model level or at the column level.

​How it works

​Default Monitors by Data Type

​Test configuration

How it works

Default Monitors by Data Type

Test configuration