In February 2017 Power BI introduced a new function called TREATAS. It simply creates virtual relationships between tables. But why we should create virtual relationships instead of physical ones? Actually, we shouldn't. It is recommended to use physical relationships whenever it's possible, but Power BI is not always sunshine and rainbows. Especially when we need to create relationships between tables that have different granularities. For this example, I have an advertising budget table for the year 2019 and a calendar table for many years:
In this situation, I need to create a relationship between my budget table and calendar table, so that I can create time intelligence measures. But I can't do it, because these two tables have different levels of granularities. While the calendar has a row for every day, the budget table has a row for every month, which means my calendar has higher granularity than my budget table. If we create relationships between tables that have different granularities, we don't get a direct error, but we get wrong results. First, I have created a relationship between the year columns of both tables and then month columns. Not surprisingly, I got the same wrong result both times.
It's not surprising that I get the wrong results because Power BI doesn't know which month belongs to which year when I have the relationship between month columns. And when I make the relationship between years, it's not able to break the total number into months because there is no relationship between month columns. So what do we do now? We need to use the TREATAS function to solve our problem. Microsoft's definition for the function is: "Applies the result of a table expression as filters to
columns from an unrelated table." As it doesn't say much as a definition, here comes my definition: "With this function, first we create a new granularity level which matches each other, and then we create a virtual relationship between tables using that new granularity level." The good thing is that all of these happen virtually and it's very easy to do. Before using the function, let's check the syntax:
TREATAS(table_expression, <column>[, <column>[,
<column>[,…]]]} )
And here comes the measure to solve the problem:
First of all, we need to use the CALCULATE function to change the filter context of our budget amount column, where we have the values for each month. Then, for the first parameter of TREATAS function, we create a virtual summary table of our calendar with the columns we need, and for the second part, we do the same thing for our Advertising Budget table by using the SUMMARIZE function. Doing this allows us to match them at the same granularity level so that a virtual relationship can be created between them. This is the table with the measure above:
Problem solved! It's amazing what this function can do and how it can save the day. To make the function work properly, we need to have the same amount of columns in the same order as the first and second parts of the parameter. Lastly, there shouldn't be any physical relationship between the two tables. If we need a physical relationship between them for other reasons, we should create it using the USERELATIONSHIP function.
Comments