Right now, the data is not ideal for analysis. Keeping in mind how I want to use the data, I need to perform some cleansing and transformation tasks. Any time I work with a new data source, I look to see if I need to do any of the following:
Remove unneeded rows or columns. Power BI stores all my data in memory when I have the PBIX file open. For optimal performance when it comes time to calculate something in a report and to minimize the overhead required for my reports, I need to get rid of anything I don’t need.
Expand lists or records. Whether I need to perform this step depends on my data source. I’ve noticed it more commonly in JSON data sources whenever there are multiple levels of nesting.
Rename columns. I prefer column names to be as short, sweet, and user friendly as possible. Short and sweet because the length of the name affects the width of the column in a report, and it drives me crazy when the name is ten miles long, but the value is an inch long—relatively speaking. User friendly is important because a report is pretty much useless if no one understands what a column value represents without consulting a data dictionary.
Rearrange columns. This step is mostly for me to look at things logically in the query editor. When the model is built, the fields in the model are listed alphabetically.
Set data types. The model uses data types to determine how to display data or how to use the data in calculations. Therefore, it’s important to get the data types set correctly in the Query Editor.
It’s a fun topic to use for learning about Power BI…says the guy wearing a Blue Jackets shirt right now…