Download presentation
Presentation is loading. Please wait.
1
1 ACCTG 6910 Building Enterprise & Business Intelligence Systems (e.bis) Olivia R. Liu Sheng, Ph.D. Emma Eccles Jones Presidential Chair of Business Dimensional Modeling: Guidelines and Common Mistakes
2
2 Business Questions Guideline #1 –Be more detail-oriented with measurements (O) How many times customers who contacted agents made purchases? (O) How much sales did customers with agent contacts generated? (X) How many customers with agent contacts made purchases?
3
3 Business Questions Guide #2: Enrich measurements and constraints using available data E.g., task time, agent page click duration, # of tasks and # of agent clicks
4
4 Guide #3: Fact Grain Not Fact table! Find it from source table first Find the lowest level possible
5
5 Fact Tables Guide #4: The attributes are not consistent with the level of grain (X) # of tasks, average task time, # of page accesses, average access time in the lowest-level-grain fact tables (O) Task time, access duration
6
6 Fact Table Guide #5: Include derived attributes only if the original attributes are measures themselves (X) # Task_key # Agent_key #Begin_time_key # End_time_key # Customer_key # Product_key. Begin_time. Exit_time # Task_key # Agent_key #Begin_time_key # End_time_key # Customer_key # Product_key. Task_time (O)
7
7 Dimensions Guide #6: Always include time dimensions, one for each different time stamp Guide #7: Data in dimensions and facts must originate from tables that can be joined in the source DB. (X) Action dimension with AgentClick Fact
8
8 Unavailable data (O) Portions of web pages accessed, customer satisfaction, direct relationships to sales and actions Guide # 7: Don’t design the star schema using unavailable data without justification
9
9 Guide #8 Don’t connect fact tables to each other! Use conformed dimensions!
10
10 Project Milestone 1 Mean: 85 90 – 100: 6 80 – 89 : 3 70 – 79 : 2
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.