Data Modeling with Microsoft Power BI

Epilogue

Congratulations—you made it through to the last page! Thank you so much for spending your precious time reading this book. I hope you enjoyed reading it as much as I enjoyed writing it. Before I let you go, here’s a quick summary of the most important things when it comes to data modeling for Power BI:

• You should sit down and understand your data and the business’s needs before connecting to the first data source. Data profiling in Power Query can help you understand the data; it shows descriptive statistics for each column.

• Bringing the source’s information into a dimensional model is crucial when it comes to Power BI. Any shortcut taken here will come back to haunt you at a later stage. (I speak from experience!) Many-to-many relationships and bi-directional filters exist only for very special use cases. I build many data models solely based on one-to-many relationships and single-directed filters.

• Make sure that every table is connected to at least one other table. Unrelated tables are useful only in special cases (like when you need a non-equi-join). If you can’t connect a table to other tables and the table is not used in connection with other tables of the semantic model, consider removing this table from this semantic model and building a different semantic model for it.

• Push all necessary transformations as early as possible up the data chain. Use Power Query over DAX for this task. If you can, convince the people providing you with Excel sheets to bring ...