Profile

Dataset profile table

The dataset profile table contains a variety of insights about column-level data quality metrics. These automatic insights bring together numerous different statistics based on the results of your DQ job to reveal the complexion of your dataset. The profile can discover attributes that help to delineate between the relative metrics around numeric versus non-numeric data.

Tip You can click anywhere in a row on the Profile page to expand it and view additional column-level insights.

Column	Description
Column	The column of your dataset whose insights the row contains.
Sensitive	View, add, or edit the sensitive label of a column. When a column has a sensitive label, a green chip with the name of the sensitive label displays. If a column does not have a sensitive label, a clickable + displays in the column. Go to the Adding Sensitive and Data Class labels section to learn how to add sensitive labels. Tip Admin users can manage sensitive labels on the Sensitive Labels Management page of the Admin Console.
Data Class	View, add, or edit the data class of a column. When a column has a data class, a green chip with the name of the data class displays. If a column does not have a data class, a clickable + displays in the column. Go to the Adding Sensitive and Data Class labels section to learn how to add data classes. Tip You can manage data classes on the Data Classes page.
Profile	Baseline (B) is the historical profile of a dataset.
	Run (R) is the profile of the current run of a dataset.
Defined Type	The data type of a column when the dataset was created.
Inferred Type	The data type(s) of a column that Collibra DQ learns are possible deviations from the defined type. If a column does not contain a potential deviation, then the data type matches the defined type.
Quality	Hover your cursor over the chart to see a percentage breakdown of the field values of the rows in a column. The represented field values are Filled, Mixed, Null, and Empty. Metric Type Description Filled [1] Integer The percentage of data that is numeric (or non-numeric) in a numeric (or non-numeric) discovered column. Mixed [String] Integer The percentage of data that is non-numeric (or numeric) in a numeric (or non-numeric) discovered column. Null [] The percentage of data that has no value at all. Empty [""] The percentage of data that has a string instance of zero length. See an example
Unique	The unique distribution, also known as cardinality, in a column. This can be an important part of understanding if the values in a column continues to be unique, or if it contains the same values. For example, the expectation of an "ID" column might be that all values should be unique, whereas the expectation of a "date" column might be that all values should be the same.
	B is the number of unique values in the baseline profile of a column.
	R is the number of unique distributions in the profile of a column from the current run.
	% is the percentage the baseline and run values deviate from each other.
Null	The percentage of data that has no value at all.
	B is the percentage of null values in the baseline profile of a column.
	R is the percentage of null values in the profile of a column from the current run.
	% is the percentage the baseline and run values deviate from each other.
Empty	The percentage of data that has a string instance of zero length.
	B is the percentage of empty values in the baseline profile of a column.
	R is the percentage of empty values in the profile of a column from the current run.
	% is the percentage the baseline and run values deviate from each other.
Mixed	The percentage of data that Collibra DQ observes as non-numeric (or numeric) in a numeric (or non-numeric) column.
	B is the percentage of mixed values in the baseline profile of a column.
	R is the percentage of mixed values in the profile of a column from the current run.
	% is the percentage the baseline and run values deviate from each other.

Column insights

Column insights become visible when you drill down into a column. These insights include key statistical data points, data visualization tools, and overviews of Adaptive Rules and Rules. Click the expand icon to the right of the column name to drill down into the column and reveal insights.

Column-level statistics

When data is available, a drill down into a column reveals four sections on the left:

• The Stats section shows high-level statistics, including:
  • Min value
  • Q1 (displays "N/A" when the data type is string or a date)
  • Median value (displays "N/A" when the data type is string or a date)
  • Mean value
  • Q3 (displays "N/A" when the data type is string or a date)
  • Max Value
  • Min string length (only displays when the data type is string)
  • Max string length (only displays when the data type is string)
  • Min precision (only displays when the data type is double, float, or decimal)
  • Max precision (only displays when the data type is double, float, or decimal)
  • Min Scale (only displays when the data type is double, float, or decimal)
  • Max Scale (only displays when the data type is double, float, or decimal)
• The Distribution section shows the number of unique values.
• The Completeness section shows the following donut charts:
  • Null percentage
  • Empty percentage
  • Mixed data type percentage
• The Business Term section shows the name of the business term and its description.
  • You can click the to the right to apply a business term to your column or edit an existing one.

Column-level data visualization

Column-level rules

The Histogram tab provides a visual histogram overview of all of the columns in your dataset where histogram data is available. From this tab, you can also see the number of distinct values that a column contains, as well as the frequency appears in a column when you hover your cursor over a histogram bar, as shown in the screenshot above.

Correlation

The correlation tab shows a graphical matrix of the correlation coefficient between two or more columns in your dataset. This can help you discover hidden column relationships in your dataset and measure their strength. Using just a few examples from the above screenshot, Collibra DQ observed the following relationships:

• There is a very strong relationship between the columns "ACTIONID" and "EFF_DATE"
• There is a strong relationship between the columns "BLOOMBERGCOMPANYID" and "CP_SH_FRACTIONAL"
• There is a moderate relationship between the columns "ANN_DATE" and "CP_RECORD_DT"

You can hover your cursor over any cell to see the correlation details between the two columns.

Note Larger datasets with many relationships between columns might not display all columns in your dataset on the matrix. However, relationships between non-visible columns will still display in the list on the right side of the matrix, when applicable.

Data Preview

The Data Preview tab shows a preview of the data in the columns in your dataset. This gives you insights into high-level statistics, such as:

• Maximum value in the column
• Mean value
• Minimum value
• Number of unique values
• Columns with labels or designations, such as Sensitive Labels and Parent Keys

You can also see columns where Collibra DQ discovered specific detections, such as the Shape observation in the "symbol" column in the example above. Additionally, you can click + Quick Rule and select either a Data Type, Data Class, or Template rule from the dropdown menu to automatically apply it to your dataset. You can then find the quick rule on the Dataset Rules page of your dataset.