Assess Data Quality
An evaluation can be carried out directly on data or remotely via questionnaires.
HOPEX Data Governance provides by default a data assessment template that focuses on the following dimensions:
• completeness
• accuracy
• consistency
• validity
• uniqueness
• freshness
Quality Dimensions
A dimension of data quality is a measurable characteristic of the data. The data quality dimensions establish data quality requirements.
HOPEX Data Governance delivers a number of predefined dimensions. To display the list of predefined dimensions:
1. Click the Compliance > Data Quality navigation menu.
2. In the edit area, click the Dimensions tab.
|
Dimension
|
Description
|
|---|---|
|
Accessibility
|
Reflects the ease of access and use of the data, at a practical level (quick, without outside intervention).
|
|
Timeless
|
Indicates the extent to which the data represents reality as of the required time. Timeliness of data implies that the data has been updated as necessary to remain relevant.
|
|
Consistency
|
Identifies the level of consistency in the data, the lack of difference when comparing two or more representations of a thing to a definition.
Example
Below is an inconsistency in the data format.  |
|
Completeness
|
Identifies the level of completeness of data and missing properties.
Example:
Below some columns have no value (in red) and others are truncated (Dupont@Samp.gm)  |
|
Confidence
|
Are data governance, data protection and data security in place? What is the reputation of the data, and is it verified or verifiable?
|
|
Accuracy
|
Identifies the level of accurate, reliable data.
Example:
Below, for Dupont, the position and the department are reversed.
For Durand, the item displays a typographical error
For Rene, the department displays an erroneous value.  |
|
Freshness
|
This criterion assesses whether the information is available at the required time.
The freshness of the data is essential to have a good view of a situation at a given time and to make decisions about the data. Freshness is important in two ways: a short delay between the data collection and its analysis and a short delay between the reporting and the resulting optimization or correction action.
|
|
Relevancy
|
Relevance of data refers to the extent to which the data meets the needs of users. Information needs may change and is important that reviews take place to ensure data collected is still relevant for decision makers.
|
|
Data Security
|
Data security covers the notion of empowerment (authorization of access to sensitive data), measures taken against the loss of information; controlling the risk of sensitive information leaks.
|
|
Traceability
|
Traceability makes it possible to follow the progress of information from its collection to its return, including its processing. Very often it is associated with the history of a process or a product.
|
|
Uniqueness
|
This criterion assesses the level of uniqueness of the data.
Example:
The "Client" table must not contain the same occurrence twice, each record must be unique.
|
|
Usability
|
Is the data understandable, simple, relevant, accessible, maintainable and at the right level of precision?
|
|
Value
|
The value of the data reflects its worth: is there a good cost/benefit ratio for the data? Are they being used optimally? Do they jeopardize the safety or privacy of individuals or the legal responsibilities of the company? Do they support or contradict the company's brand image or message?
|
|
Validity
|
Identifies the level of valid data. Data are valid when they conform to the syntax (format, type) of their definition.
Example:
The value of the "Available units" field on Prod1 should not be negative.
A withdrawal date is set to Prod2 but the field "Available units" does not display a null value.  |
|
Reasonability
|
Reasonability asks whether Data are in the correct ranges, for example check the max and mins, the Distribution and the outliers.
HOPEX provides an Excel template that allows you to evaluate the data criteria in a file and import them into your repository.
|
Evaluation Objects
An evaluation can deal with the following objects:
• concept, concept view
• class, data view
• table, physical view
• all data areas (business, logical and physical)
Direct Assessment
To directly assess a data item:
1. Open the properties of the data item in question.
2. Select the Evaluation page.
3. Click Evaluate.
4. On the page that appears, select a value for each question. See Quality Dimensions.
5. Click OK.
Assessment By Campaign
The functional administrator can create evaluation campaigns or sessions for data.
On creation of a campaign, questionnaires are sent to designated respondents to obtain qualitative estimations on the objects for which they are responsible.
For more details on campaigns and sessions, see Assessment Campaigns.
Prerequisites for Data Evaluation
Before starting a data evaluation campaign, you must first prepare the work environment. Ensure that you have defined respondents for the data, and specify for each one the entity to which he/she is attached as well as an email.
Creating assessment campaigns
To create an evaluation campaign with the template provided as standard:
1. Click the Tools > Assessment Campaigns navigation menu.
2. In the edit area, under the Campaigns tab, click New.
The campaign creation page appears.
3. Enter the name of the campaign.
4. Select Evaluation Template "Data Quality Evaluation"..
5. Modify the Calendar if required.
The calendar serves to initialize the begin and end dates of the evaluation campaign. 6. Specify the Begin Date and the End Date.
7. Click Next.
8. In the Scope Selection window, select the objects that define the evaluation context.
The context encompasses the elements of the branch that extends from the object in question up to the root.
If you deselect a node of a branch, only the child elements of this branch are deselected.9. Click Next.
10. In the preview window, click Refresh the Report.
Elements that will be assessed appear.
In particular, you can view:
• evaluated characteristics (defined in the evaluation template)
• evaluated objects
• the context objects
• evaluation nodes which correspond to objects placed in their context objects, associated with respondents.
• respondents
11. Click OK.
For more information on campaigns, see Assessment Campaigns.
See also:
