Business Analytics

 View Only

Modeling Enhancements in Cognos Analytics 11.0.4

By KEVIN MCFAUL posted Fri September 16, 2016 03:16 PM

Cognos Analytics continues to make it easier to prepare data for reports and dashboards. Here are some features related to data modules that are new in version 11.0.4.

Hiding Tables / Columns
You can now hide tables and columns in a data module. The hidden tables or columns remain fully functional and visible in the modeling interface, but they are not visible in the reporting and dashboarding interfaces.

Use this feature to provide an uncluttered view of metadata for report and dashboard users. You may want to expose a particular calculated column but not the column(s) that it references. Join key columns are good candidates to hide because they are typically not needed in reports and dashboards despite being essential for combining tables together.

Undo / Redo
While designing a data module, you can now undo and redo changes made during your session.

Only changes to the data module itself will be affected – your view of the screen won’t be disrupted unnecessarily. For example, if you change a property on a column and then open the data grid to see the results, clicking on undo will revert the property change as opposed to closing the grid even though opening the grid was the last user interaction before the undo. It would have been a lot easier for the Cognos Analytics development team to simply undo the last interaction regardless of what it may be, but that can be disorienting, so we went the extra mile because it was important for the user experience.

Clean NULL values
Business users can get confused by the difference between zeros and NULL values. In 11.0.4 we’ve extended data cleansing options with NULL-handling operations. You can replace NULL values with text, numbers, and date/time values that you specify, e.g., for a Discount Amount column you can specify 0.00 to replace the cells where the amount is unknown (NULL).

You can also specify a certain value to be replaced by NULL. For example, if you have a column that sometimes uses the value “N/A” and sometimes contains a blank (NULL), you can have them handled consistently by replacing “N/A” with NULL.

Note that when Cognos Analytics detects that a database restricts a column from containing NULL values, the NULL-handling options will not be presented on the Clean dialog of that column.

We’ve added a validation feature to give you a heads up if there’s something in your data module that won’t work in reports and dashboards. Erroneous objects in the data module are identified by red exclamation marks. Descriptions of errors and corrective actions are shown in the validation tray when it is open.

Validation will identify problems including, but not limited, to the following:
• A table or column that a data module is based on no longer exists in the source.
• A calculation expression is invalid.
• A filter references a column that no longer exists in the data module.
• A table or column that is referenced in a join no longer exists in the data module.

Recognizing Time and Locations
Cognos Analytics 11.0.4 will look for columns within a data module that contain time or geographic location data and mark them accordingly using the Represents property. You can overwrite these properties if Cognos Analytics incorrectly infers that a column contains temporal or geospatial information (perhaps a misleading column header?) or if column that does contain such information was overlooked.

The Represents property will drive smarter defaults in dashboards to get you going faster. For example, if a column is associated with countries, then a map will typically be the default visualization when that column is dropped onto the canvas. You can count on more, smarter behaviors with every new release of Cognos Analytics.

Smarter Relationship Detection
In all versions of Cognos Analytics, you can use intent-driven modeling to add tables into your data module. Intent-driven modeling proposes tables to include in the data module, based on matches between the terms you supply and information in the underlying sources.

In 11.0.4 we’ve added new smarts to intent-driven modeling to infer relationships (joins) between tables. I want to tell you all about the algorithms behind these smarts, but that would be giving away important trade secrets to our competitors!