Expected Values Check Examples

The situation: Your orders table holds one row per order. The o_orderstatus column must be one of three values: O (open), F (filled), or P (pending). Any other value is invalid and should fire a Record Anomaly so the responsible team can fix it.

Check configuration

Payload

{
    "description": "Order status must be O, F, or P",
    "rule": "expectedValues",
    "fields": ["o_orderstatus"],
    "container_id": 145,
    "coverage": 1,
    "filter": null,
    "properties": {"list": ["O", "F", "P"]},
    "tags": ["enum", "orders"],
    "additional_metadata": {"jira": "DATA-1101"},
    "anomaly_message_field": null,
    "template_id": null,
    "status": "Active",
    "owner_id": 7,
    "default_anomaly_assignee_id": 18
}

Sample Data

What gets flagged

Row 4's o_orderstatus = X is not in the list of expected values. The platform raises one Record Anomaly tied to the o_orderstatus field on that row. The other columns (o_orderkey, o_totalprice) are not part of the violation and are not highlighted.

Flowchart

graph TD
    A["No filter, evaluate all rows"] --> B{"Is o_orderstatus NULL?"}
    B -->|Yes| D["Row passes"]
    B -->|No| C{"Is o_orderstatus<br/>in (O, F, P)?"}
    C -->|Yes| D
    C -->|No| E["Emit Record Anomaly on o_orderstatus"]

Equivalent SQL

-- Rows the Expected Values check would flag on the orders table.
SELECT o_orderkey, o_orderstatus, o_totalprice
FROM orders
WHERE o_orderstatus IS NOT NULL
  AND o_orderstatus NOT IN ('O', 'F', 'P');

The situation: Your customers table is global, but the compliance team only requires the country code vocabulary to be enforced on North-American customers (region = 'NA'). The accepted codes for North America are US, CA, and MX. Customers in other regions should be ignored by this check. They are validated by region-specific checks elsewhere.

Check configuration

Payload

{
    "description": "North-American country codes must be US, CA, or MX",
    "rule": "expectedValues",
    "fields": ["country_code"],
    "container_id": 212,
    "coverage": 1,
    "filter": "region = 'NA'",
    "properties": {"list": ["US", "CA", "MX"]},
    "tags": ["compliance", "geo"],
    "additional_metadata": {"jira": "DATA-1212"},
    "anomaly_message_field": null,
    "template_id": null,
    "status": "Active",
    "owner_id": 7,
    "default_anomaly_assignee_id": 24
}

Sample Data (filtered to region = 'NA')

What gets flagged

Row 3's country_code = BR is not in the North-American vocabulary. One Record Anomaly is emitted on the country_code field of that row. The region column is part of the filter (not the rule) and is not highlighted.

Flowchart

graph TD
    A["Apply filter: region = 'NA'"] --> B{"Is country_code NULL?"}
    B -->|Yes| D["Row passes"]
    B -->|No| C{"Is country_code<br/>in (US, CA, MX)?"}
    C -->|Yes| D
    C -->|No| E["Emit Record Anomaly on country_code"]

Equivalent SQL

-- Rows the Expected Values check would flag on the customers table.
SELECT customer_id, region, country_code
FROM customers
WHERE region = 'NA'
  AND country_code IS NOT NULL
  AND country_code NOT IN ('US', 'CA', 'MX');

The situation: Your articles table has a tags column of type Array[String]. Every tag must come from the editorial vocabulary news, sports, tech, finance. A single out-of-vocabulary element in an array invalidates the row, even if the other elements are valid.

The platform auto-enables element-wise evaluation for Array[String] fields, so you do not need to set properties.is_element_context manually.

Check configuration

Payload

{
    "description": "Article tags must come from the editorial vocabulary",
    "rule": "expectedValues",
    "fields": ["tags"],
    "container_id": 318,
    "coverage": 1,
    "filter": null,
    "properties": {
        "list": ["news", "sports", "tech", "finance"],
        "is_element_context": true
    },
    "tags": ["vocabulary", "editorial"],
    "additional_metadata": {"jira": "DATA-1330"},
    "anomaly_message_field": null,
    "template_id": null,
    "status": "Active",
    "owner_id": 7,
    "default_anomaly_assignee_id": 31
}

Sample Data

What gets flagged

Row 3's tags array contains gossip, which is not in the editorial vocabulary. The whole tags cell is marked because the violation is at the field level: the array as a whole failed, not one element in isolation. Rows with an empty or NULL array pass.

Element-wise evaluation runs as a field-level check, so the result is a Shape Anomaly for the field even at 100% coverage. Unlike the scalar scenarios above, no per-row Record Anomaly is produced.

Flowchart

graph TD
    A["No filter, evaluate all rows"] --> B{"Is tags NULL or empty?"}
    B -->|Yes| D["Row passes"]
    B -->|No| C{"Is every element of tags<br/>in (news, sports, tech, finance)?"}
    C -->|Yes| D
    C -->|No| E["Count the row as failing.<br/>The field-level result is reported<br/>as one Shape Anomaly on tags."]

Equivalent SQL

-- Rows the Expected Values check would flag on the articles table.
SELECT article_id, title, tags
FROM articles
WHERE tags IS NOT NULL
  AND size(tags) > 0
  AND NOT array_forall(tags, x -> x IN ('news', 'sports', 'tech', 'finance'));