Microsoft Power BI Data Analyst Associate Study Guide

Chapter 8. Semantic Model Development

If you were to ask four-year-old me what would happen if I jumped off the roof of a house wearing a cape made from a bedsheet, I would have told you that I would fly. I had seen it demonstrated many times on television. Superman did it. Somehow, Batman pulled it off, and Spiderman didn’t even need the cape. Most four-year-olds can jump off a bed and even their brother’s top bunk without breaking a limb, but the same physics won’t work at scale.

Such is the case with data modeling. If you have a spreadsheet containing data, you can visualize some aggregate values by grouping and slicing certain columns. You might even be able to create a time series chart if you happen to have a continuous series of dates and numeric values that summarize correctly, but there are significant limitations, and some calculations and visual behaviors simply will not work. The same is true if you were to query a relational database using a SQL join statement to populate a table. You can capture the query results as a table in a Power BI report and perhaps meet your immediate report requirements, but additional calculations and visual techniques simply won’t work in the future.

I’m reminded of a business dashboard product that I investigated for a consulting client about 10 years ago. It was offered as an online service with the promise that a customer would upload their uncleansed data to the service, which would magically curate the information and then visualize ...