Hi All,
I'm modelling a FM where my QS mostly doesn't have Facts. At the maximum i can consider # of hrs leave, Over time hrs etc...
So with this kind of data source its hard for me to implement star schema grouping in the model.
Is it ok to create a model without a star schema. Also this model is a relational model.
If i can proceed without implementing star schema grouping, how do i need to go with an approach? and what are the validation do i need to do in the cardinality level and so on.
Kindly suggest me.
There is no reason why you could not build a model based on a different principle than starschema. Starschema's are required if you want to go for multi-fact queries.
As mentioned countless times before.. Star schema grouping is first and foremost a wizard to make life a little easier. (especially for presentation)
You can make life a little easier if you can define a custom fact object storing just keys and facts. It need not be an imported table, cause it is perfectly possible to define an SQL query subject for the purpose. This subject would/could act as a semi-fact.
In your case the facts do not originate from a datasource, but are generated/calculated. Nothing wrong with that. I once build a model simply around calculated time-differences, generating fact-data solely based on date/time calculations
Thank you very much for your suggestion.
As per you suggested i have created a custom fact and things are fine now.