In Data Discovery the generation of metadata is done according to some principles:
- Everything interpreted as string (text) will be added as a dimension attribute + as 2 types of measures:
distinct count measures
- Everything interpreted as a number (integer or decimal) will be added as 6 types of measures:
And also as a dimension (in the raw folder of your source tab overview in TARGIT)
- Everything interpreted as a data will generate 2 different hierarchies:
And also every part of these hierarchies will be generated as stand alone attributes in the raw folder of the source data overview
This amounts to a lot metadata - and often a lot of this is redundant for the purpose of analyzing the data in TARGIT.
To finetune this (remove a lot of it!) you edit the cube and hit the 3 horisontal lines ("hamburger menu") next to the sum symbol and pic Attribute settings:
Now you can decide exactly what should be removed from the list of metadata - look at this example:
2 Examples of "clean-up":
- The Sell-to Customer No. is numeric - and so Data Discovery has generated all the different types of measures, but you should probably untick the Avg, Min, Max and Sum for this attribute
- The Amount is a measure - and maybe you should untick this in the right-most tickbox (dimensions) - unless, of course you want to be able to filter every customer below a certain Amount)
Once you are done, you click Save and you have made life a little less confusing for the report designers, who are using this cube.