ONE Desktop ONE DQ&C ONE MDM ONE RDM ONE Runtime Server
latest (15.4.0) 15.3.0 15.2.0 15.1.0 14.5.x 13.9.x 13.8.x 13.7.x 13.6.x 13.5.0 13.4.x 13.3.x 13.2.x
User Community Service Desk Downloads
If you can't find the product or version you're looking for, visit support.ataccama.com/downloads

Create DQ Evaluation Rule

Before you start

Check the following topics:

Create new DQ evaluation rule

As well as being created from scratch, DQ evaluation rules can be created from the attribute statistics. For example, create a DQ evaluation rule for the correct format for social insurance numbers using existing masks from profiling results.

Create rule from mask

To create a rule:

  1. Navigate to Data Quality > Rules

  2. Click Create.

    Rule listing

  3. Provide the following information:

    • Name: The name of the rule.

    • Description (optional): A description of the rule and how to use it.

    • Rule definition source (optional): A descriptive name for the data source associated with the rule.

    • Owner (recommended): Select the owner of the rule. Owners are responsible for the accuracy of the information provided and need to approve any changes made to the rule before the rule is published.

    • Steward: Select the data steward of the rule. Data stewards are typically subject matter experts in a specific domain and their role is to define data quality rules and the related glossary terms.

  4. Click Save to create a draft of the rule.

    New rule missing implementation

  5. Before publishing or submitting an approval request, you need to define the rule logic consisting of at least one input attribute and condition. Continue to Rule implementation.

Rule implementation

Once you have created the rule, go to the Implementation tab

Inputs

  1. Add input attributes and select the data types.

    Input attributes are abstract at this point. The number of input attributes defined here and their rule logic become available when you assign rules to catalog items.

    The attribute names can be as you choose and do not affect the implementation.

    If special characters or reserved words are used in input names they must be enclosed in square brackets or publishing fails.
    Inputs

    1. (Optional) Assign terms to the attribute.

      Assigning terms to input attributes during implementation results in this rule being suggested by the system in the instance an attribute also has this term.

    2. (Optional) Create Variables from your inputs. Within the Variables section, you can apply a range of transformations to your input attribute. For example, apply the rule conditions only on trimmed versions of string attributes. To create variables:

      • Select a name for your variable. To add fields for additional variables, select Add Variable.

        As with the input attribute names, choose a meaningful name or leave as default.

      • Select the attributes on which the transformations are applied. Use the dropdown to select from the list of input attributes - you apply the transformations in the next step.

        Alternatively, if you are an advanced user, use the dropdown to select Advanced Expression and define the attribute and the transformations using ONE expression language.

      • Add transformations using the Add transformation option. The transformations available depend on the data type of the selected attribute: a full list of types and available transformations is provided below.

        You can apply multiple transformations to one attribute, but you can only add them in a sensical sequence. For example, if you start with an attribute of data type String and apply the transformation To integer, you will not then be able to add the Uppercase transformation, as this is only available on string.

        For the same reason, you can only remove the last transformation in a sequence at any given time. To remove the transformation, hover over the last transformation and use the x to delete.
        Expand to see all available transformations
        Datatype Available transformations Description

        String

        To float

        Changes data type from string to float.

        To integer

        Changes data type from string to integer.

        To long

        Changes data type from string to long.

        Uppercase

        Converts string characters to uppercase.

        Lowercase

        Converts string characters to lowercase.

        Trim

        Removes spaces from both ends of the string.

        Squeeze spaces

        Trims string and replaces repeated spaces with a single space.

        Remove non-digits

        Removes non-digit characters from string.

        Remove non-letters

        Removes non-letter characters from string.

        Integer

        To float

        Changes data type from integer to float.

        To date

        Changes data type from integer to date.

        To long

        Changes data type from integer to long.

        Boolean

        To String

        Changes data type from Boolean to string.

        Date

        To datetime

        Changes data type from date to datetime.

        To string

        Changes data type from date to string.

        Datetime

        To date

        Changes data type from datetime to date.

        To string

        Changes data type from datetime to string.

        Long

        To float

        Changes data type from long to float.

        To string

        Changes data type from long to string.

        To date

        Changes data type from long to date.

        To datetime

        Changes data type from long to datetime.

        Float

        To string

        Changes data type from float to string.

        Floor

        Rounds to the nearest integer that is less than or equal to float value.

        Ceiling

        Rounds to the nearest integer that is greater than or equal to float value.

        Round

        Rounds to nearest integer.

Rule logic

  1. Select the dimension that the rule should validate:

    • Validity: Validity rules verify the usability of the data (for example, regarding data format, data content or attribute relations). The default results are Valid and Invalid.

    • Uniqueness: Use this dimension to verify that there are no duplicate values and only one instance appears in the dataset. The default results are Unique, Not populated, and Not Unique.

    • Completeness: Use this dimension to verify that the value field is filled. The default results are Complete and Not complete.

    • Accuracy: Use this dimension to check whether values are accurate and reflect the true values, for example, based on reference data. The default results are Accurate, No reference available, and Not accurate.

  2. Select whether you are creating a standard Rule, Aggregation rule, or Component rule.

    Component rules need to be created first in ONE Desktop by power users. To add a component rule, select the component from the list of existing components by clicking Select a Component.

    Component rule

    The component rule needs to fulfil the following conditions in order to ensure the correct results are returned:

    • The total number of input records must match the total number of output records in the component.

    • The order of output records must match the order of input records.

    Aggregation rules require additional configuration, described in Aggregation Rules.

  3. Provide the rule conditions and an explanation of the results. There are two ways to do this: through the Condition Builder or via Advanced Expression:

    Condition builder or advanced expression

    • Select Advanced Expression to leverage ONE expressions in your rule condition (see ONE Expressions).

    • Select Condition Builder to define the rule logic using the predefined options.

If you selected Advanced Expression, enter the expression in the space provided and proceed to Test rule.

If you selected Condition Builder:

  1. Select the input attributes to use in the rule logic.

  2. Select any required modifiers, for example, Trim or Round. Different modifiers are available for string and integer inputs.

    Expand to see all available modifiers
    Modifier Description

    Value

    The given value of an attribute.

    Uppercase

    Uppercase version of string.

    Lowercase

    Lowercase version of string.

    Trim

    Removes any whitespaces from both sides of strings.

    Trim left

    Removes any leading whitespaces from strings.

    Trim right

    Removes any trailing whitespaces from strings.

    Squared

    Squares the value of an integer, float, or long data type attributes.

    Round

    Rounds the value of an integer, float, or long data type attributes to the nearest whole number.

    Average

    Averages the value of integer, float, or long data type attributes (available only when using aggregation rules and grouping).

    Min

    The minimum of string, integer, or long data types attributes (available only when using aggregation rules and grouping).

    Max

    The maximum of string, integer, or long data type attributes (available only when using aggregation rules and grouping).

    Sum

    The sum of of string, integer, or long data type attributes (available only when using aggregation rules and grouping).

  3. Select the required conditions, for example, matches_mask or has length of, and provide the necessary requirements. The options available depend on the data type of the input attributes.

    Expand to see all available conditions
    Condition Description

    Is empty

    When the input is not filled.

    This checks if the field is filled, but a field is not identified as empty if it is filled with: NULL, Null, null, ., ,, -, _, N/A, n/a, and so on.

    Is not empty

    When the input is filled.

    Is the same as

    When the input is the same as the value you define.

    Is the same as attribute

    When the attribute is the same as the attribute you select.

    Is not the same as

    When the input is not the same as the value you define.

    Is not the same as attribute

    When the attribute is not the same as the attribute you select.

    Is from the following list

    When the attribute belongs to the list you define.

    Is not from the following list

    When the attribute does not belong to the list you define.

    Has length of

    Is the same as the length you define.

    Contains

    When the input contains a substring you define.

    Is from catalog item

    When input belongs to a lookup item built from specified catalog item and attribute.

    During the rule logic definition, if you choose the option is from catalog item, a lookup item will be built automatically from the chosen catalog item and used in the rule.

    Is not from catalog item

    When input doesn’t belong to a lookup item built from specified catalog item attribute or reference data in ONE Data.

    Is from reference data

    When attribute input belongs to the lookup file you select.

    Is not from reference data

    When attribute input does not belong to the lookup file you select.

    Matches mask

    When input corresponds to the mask you define.

    Does not match mask

    When input does not correspond to the mask you define.

    Matches regexp

    When input corresponds to the regular expression you define.

    Does not match regexp

    When input does not correspond to the regular expression you define.

    Is true

    When Boolean input is true.

    Is false

    When Boolean input is false.

    Is unique

    When aggregated group is unique.

    Is not unique

    When aggregated group is not unique.

  4. Select whether the condition defined should produce the result that is, for example, Valid or Invalid. Options are different depending on the dimension chosen.

    Optionally, enter an explanation to help distinguish between multiple rules of the same dimension.

    Condition result

  5. (Optional) Assign a score to results that are then used during evaluation. See Assigning scores.

Test rule

You can optionally test the rule by clicking Test Rule and trying different inputs in the Test section. The results for each input are shown automatically in Message next to the test input row.

There is a known issue when testing component rules. However, this does not affect the functionality of the rule itself.
Expand for details

Problem There is an issue when testing component rules caused by a change in the order of results (the order of the input data was not affected). This means a component will change the order of values when producing results.

For example:

  1. You create a component rule using a uniqueness component.

  2. You select test rule

  3. You enter 1, 1, 2, 2, 3

The expected result would be that 3 is valid and 1 is invalid. Because the order of the results has been switched you will instead see that 1 is valid and 3 is invalid.

Solution If you need to test component rules it will be necessary to manually fix the defective components. This can be done by adding a shadow column with values based on the order and then sorting the results in the end with sort step.

Publish rule

After the changes are implemented, they need to be published.

Scoring records

The score is used as a numeric expression of the severity of each record’s invalidity according to the rule. This is mainly useful for monitoring projects to detect the most pressing issues and to be able to improve the data quality.

Scores are arbitrary and can be assigned to suit your project.

Typically, the logic follows that the higher the score, the worse the issue. For example, if a non-critical value is missing, the invalidity score could be 100. If a critical value is missing, the invalidity score could be 10000.

Assigning scores

It is also possible to assign scores via the validation component in ONE Desktop. This is available only for power users.

For more information, see Validation Components.

In the example shown, you have a validity rule in which an invalid record due to errors is considered of higher importance than an invalid record due to a null value, the invalid results being scored at 50 and 500 respectively.

600

Cumulative results of data quality statistics are shown on the Configuration and Results tab of a monitoring project. However, to be able to see the score of each record, it is necessary to export the project.

Next steps

Apply your newly-created DQ rule to a term. See Add Rules to Terms.

Was this page useful?