Identify expensive columns

1. Analyze the largest fact table(s) of your model to identify columns that could be excluded to improve performance.
2. For import models, search for high-cardinality columns that aren't required by the model for relationships or measures.

• In this example, the following three fact table columns are identified as candidates for exclusion from the data model

• All three columns have over 1.5M distinct values and thus don't compress well, resulting in greater data storage

1. Use the Table.RemoveColumns() function against these columns

• The size of the Power BI Desktop file is reduced from 394 MB to 227 MB
• The file is also faster to open refresh, and could support more data, while remaining below the 1 GB limit for published Power BI datasets

• The two fact tables have 27 and 24 columns, respectively, indicating that the three columns removed were among the most resource-intensive columns
• See the There's More... section for an additional method of eliminating columns