Chapter 2. A Great Foundation: AI and Data Modeling

In Power BI, the objective overall is to produce dashboards and models to support business-oriented self-service. Unfortunately, data sources are often not documented from the analysts’ perspective, so it is difficult to understand where to find the fields necessary to meet users’ needs. However, business users are creative. They can create a lot of workarounds, resulting in “shadow” IT and “shadow” data sources such as Excel workbooks that are not under the guardianship of the IT department. Often, there is limited documentation. Sometimes, the business can supply a data dictionary or a database schema on request, but it is unclear what the tables and fields mean. Frequently, data sources contain many tables, so it is hard for report writers to find the information they need.

Further, database relationships, such as table joins, are not always apparent. Database tables and views can be challenging to relate to one another and subsequently to navigate. Often, tables are difficult to understand because they are broad with many columns. For business users, this is often a productivity issue because the team cannot do the analytics they would like. Instead, the team must build and rebuild datasets in Excel or Google Sheets to get the data they need. Due to time pressures and a lack of confidence in the data sources, the team cannot take time to ask questions about the requirements. Thus, they may produce what they think is required, ...

Get Artificial Intelligence with Microsoft Power BI now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.