Forum Discussion
Multiply DirectQuery row value with imported table value (DAX)
In a composite model (mixed storage mode) I have a large fact table using DirectQuery (I am not able to change this source) and a small imported table which have a many-to-one relationship on "product_id". The fact table contains the units sold (by product_id, date, location, etc.), while the local table contains the product weight (by product_id).
What I am looking to accomplish seems quite simple:
- Multiply the units sold (fact table) for with the corresponding product weight (local table).
- Create a measure that sums all of these to a "total weight" (that can be sliced if needed).
I have looked into the calculated column approach (however functions like RELATED or LOOKUPVALUE do not work on the DirectQuery table) and using a measure. For the latter, a simple measure like SUM('fact'[units sold]) * MAX('dim Product'[weight]) works for a single "product_id" due to the relationship set, but totals and subtotals come out wrong. Adding SUMX over the 'fact' table to this for calculations at row level still seems to give me incorrect totals.
How can I workaround the restrictions of DirectQuery for this seemingly simple requirement?
Anonymous I would think the summarize doing its job, what would be the job of related function in a "normal" setting without direct query.
Summarize is doing something like what group by is doing in SQL. Its grouping/aggregating where are many product id (facttable) to the table where is only one productId towards the right weight of that productID.
Thats also why you didnt got the right totals before. You were using the max weight of the entire table instead of the right weight of the filter context.
9 Replies
Anonymous I'm not 100% sure about your mentioned data, but given that one thing did return you the right result for one single product ID, you might try the following.
If your measure is working for one single product ID like you mentioned:
Measure= SUM('fact'[units sold]) * MAX('dim Product'[weight])Since there is restrictions because of the different import variations try using x-aggregated function to get right totals:
Measure2 = sumx ('Facttable', [Measure])
Applicable88 Thanks for thinking along, I did mention in the opening post that I had tried that. However, SUMX('fact', SUM('fact'[units sold]) * MAX('dim Product'[weight])) gives for totals the units sold (total) multiplied by the maximum of all product weights. I.e. it still doesn't seem to do the calculations at row level. Does something need to be added to the measure to "force" calculations for each row, or are there limitations due to mixed storage model?
Anonymous yes. You should not put the function in, but the Measure. In every measure there there is actually a invisible "calculate" function wrapped around it. But you are calculating everything from the table. You can either try it with two measures like I explained above or you can wrap the calculate function within the "expression" part of SUMX. Like this:
SUMX('fact', calculate (SUM('fact'[units sold]) * MAX('dim Product'[weight])) )
Hope that helps.