During our migration, we have found Pivot Table Column Totals and Data Table Sum Rows are evaluated differently on Holistics. This shows up especially where aggregated data model records are applicable to multiple selected dimensions in the table / pivot table.
I understand improvements to Data Tables are in consideration and development. It may be beneficial to leverage a similar method for Sum Rows as is currently implemented in Pivot Table Column Totals. An example and details below:
For example, in a Dataset with the following model relationships (one < many)
Inventory < Customer Contacts > Inventory Customer Details (sales)
Customer Contacts > Customers
Customer Contacts > Contact Method (email, social)
If we build the a table with the following fields, Holistics accurately evaluates the relationships to ensure only unique records from Inventory Customer Details are evaluated.
- Customer (Steve)
- Sum of sales (1000)
If we add Contact Method, Holistics correctly evaluates the relationships to generate accurate sums within both visualization types. Consider the Customer may have been contacted about the same Inventory via multiple Contact Methods. As a result, a single Inventory Customer Details record may be referenced multiple times.
- Customer (Steve, Steve)
- Contact Method (email, social)
- Sum of sales (900, 300)
The two visualizations diverge when Column Totals and Sum Rows are applied. Pivot Table Column Totals evaluates unique records from Inventory Customer Details. As a result, the total is accurate (1000) , but does not match the visually apparent sum of the values for each Customer and Contact Method. Data Table Sum Rows is a direct sum calculated at visualization (1200), without consideration for records referenced multiple times due to the data model relationships. As a result, the total visually matches, but is inaccurate for the level of detail of the total.
