data warehouse - Is it possible to have conditional OLAP dimension aggregators? -
i have set of olap cubes, in form of snow-flake schema, each representing 1 factory.
i have 3 concepts factories behave 3 dimensions, , other factories behave 2 dimensions.
the concepts same: "products", "sales agents" , "customers".
but cases, doubt if should model purely 3 dimensional cube or should play around tweak or trick 2 dimensional cube.
cases , b ones clear me, , case c 1 generates wonderings.
case a: 3 dimensional cube
any agent can sell product company. several agents resposible same set of customers.
i model case this:

case b: 2 dimensional cube
every agent 'responsible' portfolio of customers, , can sell product customers. analysis made on 'current responsability on portfolio' if agent leaves company, customers reassigned new agent , customer uniquely belongs new agent.
i model case this:

case c: doubts
a customer may have been assigned single agent or set of several agents each 1 being responsible productcategory.
for example:
alicemanagestablesandwoods ltd,greenforest ltd.bobmanageschairs ltd,fastwheels ltd.carolmanagesforniture ltdproducttype = 'machinery', managesfrozenbottles ltdtype of product.davemanagesforniture ltdproducttype = 'consumables', manageshighceilings ltdtype of product.
question:
in example "case c":
are customer , agent independent dimensions because forniture ltd has relation both caroland dave, 3d cube?
or 2d cube, agent not independent dimension, aggregator of customer "conditioned" somehow productcategory product aggregator?
i see how model this. in advance.
here how model it:
your fact table sales.
your dimensions (probably) date, product, customer , agent. closest case a.
collapse snowflake (white entities) dimensions. presence of these entities suggest should consider whether type-2 changing dimensions needed at-time analysis.
consider bridge table capture many-to-many relationship between agent , product.
Comments
Post a Comment