Forum Discussion

mrLeyshock's avatar
mrLeyshock
Regular Visitor
5 years ago
Solved

Calculate Percent Change Using Non-Continuous Dates

I am trying to create a calculation to show percent change over time using a date column that is not a continuous date. The data shows total file sizes each time a report is run, but the report can be run on demand so attempts to use functions like PREVIOUSDAY don't really work for me because there are missing days. 

 

DataUsage

SizeReportDate
10010/1/2021
5010/4/2021
2510/7/2021
7510/15/2021
5010/17/2021
12510/18/2021
15010/20/2021
10010/22/2021

 

Given the information above, can anyone tell me the best way to create a table similar to this?

 

ReportDateSizeChange (Current Size - Previous Size)

Percent Change

(Change / Previous Size)

10/1/2021100  
10/4/202150-50-50%
10/7/202125-25-50%
10/15/20217550200%
10/17/202150-25-33%
10/18/202112575150%
10/20/20211502520%
10/22/2021100-50-33%

 

 

 

 

For background, I've attempted a variety ways to get at this calculation without success which I'll outline below.

 

I created a dates table and a previous date column:

 

DatesLoaded = SUMMARIZE(
ALL(DataUsage[ReportDate]),
DataUsage[ReportDate])
 
Row Index Column =
CALCULATE(
COUNTROWS(DatesLoaded),
FILTER(All(DatesLoaded), DatesLoaded [ReportDate] <= EARLIER(DatesLoaded [ReportDate]))
)
 
PreviousDate =
LOOKUPVALUE('DatesLoaded'[ReportDate],'DatesLoaded'[Row Index Column], 'DatesLoaded'[Row Index Column]-1)
 
I created a one to many relationship from DatesLoaded to DataUsage on ReportDate:
 

 

Then I created the following measure:
Previous Size = CALCULATE(SUM(DataUsage[Size]), FILTER(DataUsage, DataUsage[ReportDate]=RELATED(DatesLoaded[PreviousDate])))
 
But I don't get any results in the Previous size measure. 
 
I also tried:
Previous Size = CALCULATE(SUM(DataUsage[Size]), ALLSELECTED(DatesLoadedStatic[PreviousDate]))
 
Which returned only the current size, not the previous.
 
 
If I can get the previous size value to work correctly, I believe the fillowing would take care of the difference and % change columns:
 
Usage Difference =
VAR _CurrentDayUsage = SUM(DataUsage[Size])
VAR _PreviousDayUsage = DataUsage[Previous Size]
VAR _Result = _CurrentDayUsage - _PreviousDayUsage
RETURN _Result
 
Usage Growth = DIVIDE ([Usage Difference], [Previous Size], 0)
Need Help
  • v-yalanwu-msft's avatar
    v-yalanwu-msft
    5 years ago

    Hi， mrLeyshock ；

    You could modify the measure as follows:

    Percent Change = 
var _pre=MAXX(FILTER(ALL('Table'),[ReportDate]= CALCULATE(MAX([ReportDate]),FILTER(ALL('Table'),[ReportDate]<MAX('Table'[ReportDate])))),[Size])
return DIVIDE(MAX([Size])- _pre,_pre)

    The final output is shown below:

    Best Regards,
    Community Support Team_ Yalan Wu
    If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

4 Replies

  • HotChilli's avatar
    HotChilli
    Icon for Community Champion rankCommunity Champion
    5 years ago

    Here's previous size as a column.

    PreviousValue = 
VAR _ReportDate = DataUsage[ReportDate]
  VAR _prevDate =  CALCULATE(MAX(DataUsage[ReportDate]), FILTER(DataUsage,DataUsage[ReportDate] < _ReportDate))
RETURN
CALCULATE(SUM(DataUsage[Size]), FILTER(DataUsage,DataUsage[ReportDate] = _prevDate))

    Is that enough to get you started?

  • v-yalanwu-msft's avatar
    v-yalanwu-msft
    Icon for Community Support rankCommunity Support
    5 years ago

    Hi， mrLeyshock ；

    You could modify the measure as follows:

    Percent Change = 
var _pre=MAXX(FILTER(ALL('Table'),[ReportDate]= CALCULATE(MAX([ReportDate]),FILTER(ALL('Table'),[ReportDate]<MAX('Table'[ReportDate])))),[Size])
return DIVIDE(MAX([Size])- _pre,_pre)

    The final output is shown below:

    Best Regards,
    Community Support Team_ Yalan Wu
    If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

    • mrLeyshock's avatar
      mrLeyshock
      Regular Visitor
      5 years ago

      Thank you for sharing this. It was exactly what I needed. 

      • mrLeyshock's avatar
        mrLeyshock
        Regular Visitor
        5 years ago

        To close the loop in this. The example that you provided was perfect for the data sample that I shared. Of course my data was a bit more complex, rather than one row per date with a size value, I actually have many smaller size values for each date that I needed to sum. I was ale to modify the example to make it work.


        --CREATED AS A COLUMN

        PreviousDate =
        VAR _ReportDate = DataUsage[ReportDate]
             VAR _prevDate = CALCULATE(MAX(DataUsage[ReportDate]), FILTER(DataUsage,DataUsage[ReportDate] < _ReportDate))
        RETURN
        _prevDate


        --MEASURES


        PreviousSize =
        SUMX (
             FILTER (
                  ALL ( 'DataUsage' ),
                  [ReportDate]
                       = CALCULATE (
                            MAX ( [ReportDate] ),
                            FILTER (
                                 ALL ( 'DataUsage' ),
                                 [ReportDate] < MAX ( 'DataUsage'[ReportDate] )
                            )
                       )
             ),
             [Size]
        )

        PercentChange = DIVIDE([CurrentSize]-[PreviousSize], [PreviousSize])