Sorry Ian. I made a mistake testing the datamodule. To test I used a filter on a measure to get only a few rows,but this filter uses after auto aggregation.
This was the reason for the unexpected results.
It is working fine now. The cardinality is 1:n with the n-side on the dataset. The dimension grain is the same as the fact grain. So no double counting.
------------------------------
Thomas van der Meer
------------------------------
Original Message:
Sent: Wed May 06, 2020 03:09 PM
From: IAN HENDERSON
Subject: Incorrect result when Dataset is joined to more than 1 table
By incorrect results, do you mean double counting?
What grains of detail are the dimension tables relative to your fact table (which is your dataset if I understand correctly)?
What keys are you using? What's the cardinality of the relationships?
If the dimension grain is lower than the fact grain you can get double counting if you have not specified column dependency.
------------------------------
IAN HENDERSON
Original Message:
Sent: Wed May 06, 2020 01:01 PM
From: Thomas van der Meer
Subject: Incorrect result when Dataset is joined to more than 1 table
I have some issues with datamodule which consist of a dataset joined to more than 1 other table.
I have created a dataset with some aggregated data which I use like a kind of fact table. Then I create a datamodule with this dataset and join this with 2 other (dimension) tables. These 2 other tables come from a data module with a dataserver as a source.
When I use this data module in a report then it gives incorrect results in the measure values when I select both 1 measure and a column from another table. But when the dataset is joined on only 1 table then the results are correct.
Can anybody confirm this behaviour? (I'm using version 11.1.2)
------------------------------
Thomas van der Meer
------------------------------
#CognosAnalyticswithWatson