BIchick

 Profile Answers by BIchick 

• Feb 5th, 2010

 

This is simple layout without stocking/restocking data, also without GL, specials handling, etc...
You would have to have these dimensions:

Date (including, excluding time - based on how grainual you need to go)
date key, date, date attributes - whatever you need to analyze
20100101, 01/01/2010, Monday, 1 QT

Buyer (or visitor)
buyer key, buyer attribute - whatever you like to keep
123, John Smith, 123 Main Street, New York, NY, male, 35 y.o., married...

Products
product key, product title, price,  product attributes - whatever you have
1234, stapler, $1.00, black, (made by) USRT, 45inch

and your Sales fact table:
datekey, buyer key and product key
20100101, 123, 1234

Ramanagoud

• Mar 17th, 2013

 

Dimensional model will be having original tables and will be joined directly to other tables using primary key. As in ER modelling we will not do any normalization here. We don't bother about any memory reduction by doing normalization and all. Updating tables in dimensional model is very is as compared to ER tables.