Measure to sum another measure problem

[johnfa]

Hi there

I am new to DAX and having problems calculating a measure to sum another measure and was hoping someone could point me in the right direction.  I have simplified the data below to show the problem. 

Date	Period	Product	Sales
01/01/2019	1	A	0
01/01/2019	2	A	0
01/01/2019	3	A	0
01/01/2019	1	B	3
01/01/2019	2	B	1
01/01/2019	3	B	3
01/01/2019	1	C	0
01/01/2019	2	C	1
01/01/2019	3	C	2
02/01/2019	1	A	0
02/01/2019	2	A	0
02/01/2019	3	A	0
02/01/2019	1	B	0
02/01/2019	2	B	0
02/01/2019	3	B	0
02/01/2019	1	C	0
02/01/2019	2	C	0
02/01/2019	3	C	0
03/01/2019	1	A	1
03/01/2019	2	A	0
03/01/2019	3	A	3
03/01/2019	1	B	2
03/01/2019	2	B	2
03/01/2019	3	B	0
03/01/2019	1	C	1
03/01/2019	2	C	3
03/01/2019	3	C	2

I  am trying to get a count of the number of products without any sales by date  so my output would look like this 

Date	#Products With No Sales
01/01/2019	1
02/01/2019	2
03/01/2019	0

I can create a measure on my Sales Data table that groups things by date and product returning 1 if there are no sales and 0 if there are sales for the day.

Missing Product Sales For Day =
IF (
CALCULATE (
SUM ( 'Sales'[Sales]),
ALL ( Sales[Period])
) = 0,
1,
0
)

I was then thinking i would just be able to sum these values up by day in another measure but i cannot reference the first measure in any subsequent measure without getting the following error.   

Column 'Missing Product Sales For Day' in table 'Sales' cannot be found or may not be used in this expression. 

What would be the simplest way to do this ?  I think i can do it by creating more tables using Summarize but that seems a bit clunky

many thanks
Johnfa

[YJ · Accepted solution]

Hi ,

Under Modeling , New Table put this in:

 Tb_Products_w_no_Sales = SUMMARIZECOLUMNS(Table1[Date ],

(Filter (

SUMMARIZECOLUMNS(Table1[Date ],Table1[Product],"ZERO_S",IF(SUMX(Table1,Table1[Sales])=0,0,1)),[ZERO_S]=0

)

),"Products_w_no_Sales",DISTINCTCOUNT(Table1[Product])

)

*overall idea, ZERO_S will give 0 for rows of products with no sales, and after do a distinct count on product
where Table1 is your data.

This should be what you want. i believe 

02/01/2019

2

is not correct, it should be 3. but let me know.

If you need a 0 on the missing dates, link it to a date table or left join it.

Regards

[v-frfei-msft · Accepted solution]

Hi @johnfa ,

We can create a calcualted column and new a measure based on it.

per product = 
CALCULATE (
    SUM ( 'Table'[Sales] ),
    FILTER (
        'Table',
        'Table'[Date ] = EARLIER ( 'Table'[Date ] )
            && 'Table'[Product] = EARLIER ( 'Table'[Product] )
    )
)

Measure = CALCULATE(DISTINCTCOUNT('Table'[Product]),FILTER('Table','Table'[per product] = 0))

[YJ · Accepted solution]

Hi ,

Under Modeling , New Table put this in:

 Tb_Products_w_no_Sales = SUMMARIZECOLUMNS(Table1[Date ],

(Filter (

SUMMARIZECOLUMNS(Table1[Date ],Table1[Product],"ZERO_S",IF(SUMX(Table1,Table1[Sales])=0,0,1)),[ZERO_S]=0

)

),"Products_w_no_Sales",DISTINCTCOUNT(Table1[Product])

)

*overall idea, ZERO_S will give 0 for rows of products with no sales, and after do a distinct count on product
where Table1 is your data.

This should be what you want. i believe 

02/01/2019

2

is not correct, it should be 3. but let me know.

If you need a 0 on the missing dates, link it to a date table or left join it.

Regards

[johnfa]

Thanks great that works - though not entirely sure what is happening ! Presumably the inner SUMMARIZECOLUMNS table gets evaluated first then the other columns are referring to the values returned from the inner table and not the base table Table1 ?

ie  "Products_w_no_Sales" DISTINCTCOUNT(Table1[Product]) is working on the rows returned from the inner SUMMARIZECOLUMNS table or have i misunderstood ? .  If so it would be nice if you could rename the columns to reflect where they are coming from but this does not seem possible ?.  The column "ZERO_S" for example cannot be referenced outside the inner SUMMARIZECOLUMNS table.

[YJ]

Thanks @johnfa for accepting it as the solution and the kudo,

its ready great to try to understand what is happening, so lets try:

Presumably the inner SUMMARIZECOLUMNS table gets evaluated first then the other columns are referring to the values returned from the inner table and not the base table Table1 ? YES 

ie  "Products_w_no_Sales" DISTINCTCOUNT(Table1[Product]) is working on the rows returned from the inner SUMMARIZECOLUMNS table or have i misunderstood ? . YES , it is doing a distinct count on column of base table Table1[Product] after the inner summarise after the filter. Then we assign this new(final) agregation (which would be show to the users) as "Products_w_no_Sales". if you change "Products_w_no_Sales" to "PPP" you see "PPP" in your final table field.
https://docs.microsoft.com/en-us/dax/summarizecolumns-function-dax

groupBy_columnName: Table1[Date] ( thats why finally we see the rows for all our grouped(unique) dates)

filterTable: the chuck of prehaps confusion which is your basetable after a 1st summarise follow by a filter

name: "Products_w_no_Sales"

expression: DISTINCTCOUNT(Table1[Product]

The column "ZERO_S" for example cannot be referenced outside the inner SUMMARIZECOLUMNS table. It can be referenced outside the inner SUMMARIZECOLUMNS table but not outside the Tb_Products_w_no_Sales table(afaik)

Minus the issue of been chucky with a couple of uncessary tables, you can break it down in steps for understanding. And in this way if you need to refer to it outside the table, Example ZERO_S can now be refered elsewhere

TB_P1xx: Which in my opinion is the crucial part, it identifies wether each group have zero summation of the sales. the IF here is not absolute necessary, its more of to go along with your concept of "groups things by date and product returning 1 if there are no sales and 0 if there are sales for the day." although i just realised that i did the opposite ....=)

TB_P2xx: Filter only those with no sales

TB_P3xx: Count how many have no sales in each date group.

Hope it helps, let me know if you have more question or need the PBIX.

Regards

[johnfa]

Thanks for taking the time to explain it fully thats really helpful. Being new to DAX it is certainly easier to follow if every step is broken down explicitly into a separate table although obviously a bit long winded.   It is frustrating that you cant refer to columns created in the inner table in the outer table in the same block ie "Products_w_no_Sales" -  I am used to working with SQL where you can easily create aliases for datasets and columns and refer to them later.   Would it be possible to assign the inner table to a variable and then use that later to refer to all the columns we want ?

[YJ]

Hi, @Jo 

afaik, its is not yet possible to declare global var in DAX:

https://community.powerbi.com/t5/Developer/Is-it-possible-to-have-global-variable-for-whole-report/t...

However for improve readaiblity, its possible to have  local var:

Tb_Products_w_no_Sales =

Var

InnerTable= SUMMARIZECOLUMNS(Table1[Date ],Table1[Product],"ZERO_S",IF(SUMX(Table1,Table1[Sales])=0,0,1)

Return

SUMMARIZECOLUMNS(Table1[Date ],

(Filter (InnerTable),[ZERO_S]=0)),

"Products_w_no_Sales",

DISTINCTCOUNT(Table1[Product])

)

Of course with the indentation in powerbi it should look clearer there.( i am sorry that i didnt indent it well )

I understand how you feel , for something similiar to SQL, (use VAR as above)

WITH MYTABLE 

AS

() 

SELECT *

FROM BASETABLE

but for alias , SELECT * from BASETABLE as ALIAS_MYTABLE, i think there is nothing  similiar...

My personal preference when working with Powerbi and SQL is to have a view or table function on the SQL server and have powerbi query from that to do measures, YTY , simple agregation and reporting. But i am sure this is subjective 😃

regards