Forum Discussion
Using DATEADD() and cross-filtering
Hi All !
we have 2 simple datasets + Calendar:
our screen:
Our steps:
1. choice Products in slicer Prod or leave it unselected
2. choice month in slicer Month (one or more)
3. Measure [back_month] - contains the number of selected months
back_month = (DATEDIFF(MIN('calendar'[Date]),MAX('calendar'[Date]),MONTH)+1)*-1
4. Measure [count_filter] - contains the total number of filtered rows in the txt-table for the selected months (point 2) and Products (point 1)
5. The problem in the [count_back].
Measure [count_back] - must containts the total number filtered rows in the txt-table for [back_month] months ago.
count_back =
VAR back_month = (DATEDIFF(MIN('calendar'[Date]),MAX('calendar'[Date]),MONTH)+1)*-1
VAR count_back=CALCULATE([count_filter],DATEADD('calendar'[Date],back_month,MONTH),ALL())
RETURN count_back
You can see the expected date-intervals for filtering [count_back] in Measures [date_backFr] & [date_backLs]
date_backFr = FIRSTDATE(DATEADD('calendar'[Date],[back_month],MONTH))
date_backLs = LASTDATE(DATEADD('calendar'[Date],[back_month],MONTH))
For example,
if we chose the Month interval: (2021-03) & (2021-04), then in the Measure [count_back] we must calculated the number of rows in txt for the previous month-intervals (-2 months) => (2021-01) & (2021-02)
if we chose the Month interval: (2021-03), then in the Measure [count_back] we must calculated the number of rows in txt for the previous month-interval (-1 month) => (2021-02)
At now, the measure [count_back] work correctly only if we do not use the Prod filter.
If we turn Propuct filter is on, then everything breaks down ((
How to use additional filtering by Prod in this case?
PBIX file: http://ftp.kantartns.ru/download/637ycdxjeb216eym2vw72b13q0jf73/tt.zip
I was modify my formula (using PARALLELPERIOD ) and it worked!
Nprev = VAR monthBack = (DATEDIFF(MIN('rsp'[date]),MAX('rsp'[date]),MONTH)+1)*-1 VAR xStart = STARTOFMONTH(PARALLELPERIOD(rsp[date],monthBack,MONTH)) VAR xEnd = ENDOFMONTH(PARALLELPERIOD(rsp[date],monthBack,MONTH)) VAR cntx = CALCULATE(COUNTROWS(txt),FILTER(ALL(rsp[date],rsp[month]),rsp[date]>=xStart && rsp[date]<=xEnd)) RETURN cntx
3 Replies
StenX , Try like
same period based on date range
Last Period =
var _max =maxx(date,date[date])
var _min =maxx(date,date[date])
var datediff1 = datediff(_min,_max,day)
var _maxX = _max-datediff1
var _minX = _min -datediff1
return
CALCULATE(SUM(Sales[Sales Amount]),filter(all(date),date[date]<=_maxX &&date[date]>=_minX)))Thank you for your quick response!
My apologies for the delay in reply. Our office was closed for the weekend ((I applied your idea to calculate for a displacement of the full month(s).
count_MonthBack = VAR xDiffMonthBack = (DATEDIFF(MIN('calendar'[Date]),MAX('calendar'[Date]),MONTH)+1)*-1 VAR xStartBack = STARTOFMONTH(DATEADD('calendar'[Date],xDiffMonthBack,MONTH)) VAR xEndBack = ENDOFMONTH(DATEADD('calendar'[Date],xDiffMonthBack,MONTH)) VAR xCountBack=CALCULATE([cnt],FILTER(all(rsp),rsp[date]>=xStartBack && rsp[date]<=xEndBack)) RETURN xCountBack
Unfortunately, I couldn't do it. If we use the slicer Prod, everything breaks.
Perhaps the problem is in the DATEADD(). How can this be fixed for month metrics?For check:
if we select the month 2021-04, then the date of the week should be filtered by 01.03.2021-31.03.2021 (one month)
if we select the month 2021-04&2021-03, then the date of the week should be filtered by 01.01.2021-28.02.2021 (two months)
Regards,
I was modify my formula (using PARALLELPERIOD ) and it worked!
Nprev = VAR monthBack = (DATEDIFF(MIN('rsp'[date]),MAX('rsp'[date]),MONTH)+1)*-1 VAR xStart = STARTOFMONTH(PARALLELPERIOD(rsp[date],monthBack,MONTH)) VAR xEnd = ENDOFMONTH(PARALLELPERIOD(rsp[date],monthBack,MONTH)) VAR cntx = CALCULATE(COUNTROWS(txt),FILTER(ALL(rsp[date],rsp[month]),rsp[date]>=xStart && rsp[date]<=xEnd)) RETURN cntx
