Column dependencies in data modules

Updated: Sep 27, 2019

As we consider moving from Framework Manager packages to data modules for our metadata needs, one of the critical things we need to have access to is the ever-confusing determinants that we have in Framework Manager. Without determinants, we may see double-counting or incorrect aggregating in our reports and dashboards. Until this functionality was added in 11.1, it was going to be tricky to fully move to data modules when reporting against a data warehouse. In data modules, determinants are called column dependencies. We need column dependencies for a couple of reasons. 1) When we have multiple fact tables with conformed dimensions joined to dimensions on columns that are at different grains of detail. For example, the Sales target fact and the Sales fact join to conformed dimensions on different fields – Sales target joins on the month field in the Time table, Sales target joins to Products table on Product type, and Sales target joins to Retailers table on Retailer whereas Sales fact joins on Date, Product, and Retailer site (all at a lower grain of detail).

2) When a table contains previously aggregated data. For example, a table that contains pre-summed columns for revenue at the monthly, quarterly, and yearly levels. If we add those previously summed columns to a report, Cognos may aggregate them again.

To learn more about each type of column dependencies, see the following: Column dependencies with multi-fact queriesColumn dependencies with previously summed columns As we consider moving from Framework Manager packages to data modules for our metadata needs, one of the critical things we need to have access to is the ever-confusing determinants that we have in Framework Manager.

Without determinants, we may see double-counting or incorrect aggregating in our reports and dashboards. Until this functionality was added in 11.1, it was going to be tricky to fully move to data modules when reporting against a data warehouse.

In data modules, determinants are called column dependencies. We need column dependencies for a couple of reasons.

1) When we have multiple fact tables with conformed dimensions joined to dimensions on columns that are at different grains of detail.

For example, the Sales target fact and the Sales fact join to conformed dimensions on different fields – Sales target joins on the month field in the Time table, Sales target joins to Products table on Product type, and Sales target joins to Retailers table on Retailer whereas Sales fact joins on Date, Product, and Retailer site (all at a lower grain of detail).

2) When a table contains previously aggregated data. For example, a table that contains pre-summed columns for revenue at the monthly, quarterly, and yearly levels. If we add those previously summed columns to a report, Cognos may aggregate them again.




To learn more about each type of column dependencies, see the following:

Column dependencies with multi-fact queries

Column dependencies with previously summed columns


©2019 by Atlantic Analytics. Proudly created with Wix.com