All analytical expressions ranging from simple sums and averages to custom, complex statistical analyses are implemented within DAX measures. Most measure expressions will reference and aggregate the numeric columns of fact tables, which are hidden from the Report View, as we have seen in per the previous chapter. Additional DAX measures can include filtering conditions which supplement or override any filters applied in Power BI reports, such as the net sales amount for first-year customers only. Measures can also evaluate text columns from dimension tables, such as the count of states or provinces with sales and return text and date values.

Just like the M query language, DAX is a rich, functional language that supports variables and external expression references. Multiple variables can be defined within a DAX measure to improve readability, and the results of other measures can be referenced as well, such as the Plan Grain Status measure in Chapter 3, Designing Import and DirectQuery Data Models. These layers of abstraction and the built-in code editing features of Power BI Desktop including IntelliSense and colorization, enabling dataset designers to embed powerful yet sustainable logic into datasets.