MIS 451 Building Business Intelligence Systems Logical Design (3) – Design Multiple-fact Dimensional Model
2
3 Design Steps Step 1: Decide facts and fact measurements: What do customers want to analyze? --- Facts Find measurements of fact in the source database – Fact Measurements
4 Design Steps Step 2: Decide grain of facts: Grain is the level of detail at which fact measurements will be stored in data warehouse. Data warehouse usually stores facts at their lowest level of detail.
5 Design Steps Step 3: Decide fact tables: Facts Fact Tables Fact Measurements Fact Table Attributes
6 Design Steps Step 4: Determine Primary Keys of Fact Tables Find the natural keys that can uniquely identify records in fact tables from the source database – primary key of fact table Transform natural keys into warehouse keys
7 Design Steps Step 5: Decide key dimension tables for a fact table: Key dimension tables provide primary key for a fact table. Step 6: Decide analysis dimension tables for a fact table: Analysis dimension tables provide foreign keys for fact tables. Step 7: Connect separate dimension models using conformed dimension tables A conformed dimension table is a dimension table shared by fact tables. It can be used to link separate dimensional models into a coherent one.