A fundamental knowledge for authors of DAX measures is the filter context in which the calculation logic of DAX measures is executed. The filter context is the set of rows for each table in the data model which are available to be evaluated for each value displayed in report visualizations. For example, each value of a Power BI chart that analyzes a sales amount measured by a Product Category column will usually be unique because a different set of rows (filter context) of the sales fact table was available to the measure when calculating each value.
In the following Power BI report, five distinct filters representing five dimension tables have been applied to a matrix visual of the Internet Gross Sales measure:
Filters applied to the Promotion, Date, Product, Customer, and Sales Territory dimension tables all flow across their relationships to filter the Internet Sales fact table. The Internet Gross Sales measure is a simple SUMX() expression described in the base measures and is thus evaluated against the Internet Sales rows remaining from these filters.
Each individual value in Power BI reports is computed independently. For example, the $315,028 subtotal value is not filtered by the Product Category column like other values in the matrix, and it's not calculated as the sum of the three cells in the Married column. This value is computed using the customer marital status column value of Married and the other filters from the report. See the following Measure evaluation process section for details on the DAX engine's execution process.