To convert MTD data to YTD but with 3 variables

[Catherine84]

Hi all,

I will like to seek your kind help. I am pretty new to power BI and running into trouble getting the correct formulas.

I have a set of data below. Q1 numbers are correct. Q2,Q3 and Q4 are all YTD numbers and i need to create another columns to get the it to reflect QTD numbers

I have been trying but its not working, not sure if you all can help. I have quite a few items that must consider Ie RU, account code, year. Not sure how to gel them! 

Value2 = VAR pq = IF ( [Period] = "Q1", "Q2", "Q1" ) RETURN IF ( [Period] = "Q1", [Amount], [Amount] - CALCULATE ( MAX ( [Amount] ), FILTER ( ALL ( table ), [Year] = EARLIER ( [Year] ) && [Period] = pq), Filter([Table],[RU n  Account code]=[RU]&[Account code]) ) )

Thanks in Advance

RU	Account code	Period	Year	YTD Amount
XYZ Co	Salaries and Wages	Q1	2017	20
XYZ Co	Salaries and Wages	Q2	2017	30
XYZ Co	Salaries and Wages	Q3	2017	45
XYZ Co	Salaries and Wages	Q4	2017	80
XYZ Co	Salaries and Wages	Q1	2016	30
XYZ Co	Salaries and Wages	Q2	2016	50
XYZ Co	Salaries and Wages	Q3	2016	55
XYZ Co	Salaries and Wages	Q4	2016	80
ABC Co	Salaries and Wages	Q1	2017	34
ABC Co	Salaries and Wages	Q2	2017	60
ABC Co	Salaries and Wages	Q3	2017	75
ABC Co	Salaries and Wages	Q4	2017	100
ABC Co	Salaries and Wages	Q1	2016	20
ABC Co	Salaries and Wages	Q2	2016	43
ABC Co	Salaries and Wages	Q3	2016	54
ABC Co	Salaries and Wages	Q4	2016	80
XYZ Co	Bonus	Q1	2017	10
XYZ Co	Bonus	Q2	2017	20
XYZ Co	Bonus	Q3	2017	24
XYZ Co	Bonus	Q4	2017	32
XYZ Co	Bonus	Q1	2016	3
XYZ Co	Bonus	Q2	2016	5
XYZ Co	Bonus	Q3	2016	10
XYZ Co	Bonus	Q4	2016	15
XYZ Co	Bonus	Q1	2017	30
XYZ Co	Bonus	Q2	2017	43
XYZ Co	Bonus	Q3	2017	60
XYZ Co	Bonus	Q4	2017	75
XYZ Co	Bonus	Q1	2016	12
XYZ Co	Bonus	Q2	2016	25
XYZ Co	Bonus	Q3	2016	56
XYZ Co	Bonus	Q4	2016	76

[Sean · Accepted solution]

Hi @Catherine84 ,

I would suggest shaping the data better first In the Query Editor

1) select the Period column - Transform tab - click Extract - select Last Characters - type 1 - OK

2) change the Data Type of the Period column to Whole Number

3) click Group By - select Advanced

4) add these 4 groupings - RU - Account Code - Year - Period

5) for New column name - type - YTD Amount

6) for Operation - select - Sum

7) for Column - select - YTD Amount - OK

😎 Close and Apply

9) Finally create a New Column (this would be a DAX column)

DAX Amount =
VAR Company = Table1[RU]
VAR Code = Table1[Account code]
VAR CYear = Table1[Year]
VAR CQuarter = Table1[Period] // Period column is Number (1, 2, 3, 4) representing Quarters
RETURN
    [YTD Amount]
        - CALCULATE (
            SUM ( Table1[YTD Amount] ),
            FILTER (
                Table1,
                Table1[RU] = Company
                    && Table1[Account code] = Code
                    && Table1[Year] = CYear
                    && Table1[Period] = CQuarter - 1
            )
        )

DAX Amount V2 =
VAR Company = Table1[RU]
VAR Code = Table1[Account code]
VAR CYear = Table1[Year]
VAR CQuarter = Table1[Quarter] // Quarter column is Text (Q1, Q2, Q3, Q4)
RETURN
    Table1[YTD Amount]
        - CALCULATE (
            MAX ( Table1[YTD Amount] ), //Have to use MAX!
            FILTER (
                Table1,
                Table1[RU] = Company
                    && Table1[Account code] = Code
                    && Table1[Year] = CYear
                    && Table1[Quarter] < CQuarter
            )
        )

That should do it!

Now you data is shaped better and this new Amount column will show you the total per each Quarter only.

I'm sure ImkeF can probably do my last DAX step with M as well

[Ashish_Mathur]

Hi,

So the result you want for the first 4 rows are 20,10,15,35.  Am i correct?  Also, do you want this to be a measure or a calculated column?

[Catherine84]

Yes Ashish. Thanks for replying. Yes correct I need the first 4 rows to be 20,10,15,35.. It will be great if this can be a measure!

[Ashish_Mathur]

Hi,

I have made the following assumptions:

1. There is a Calendar Table
2. There is a relationship from the Date column of your base data table to the Date column of your Calednar Table
3. In the Calendar Table, you also have the Period and Year columns
4. The Period and year columns in your visual have been dragged from the Calendar Table
5. The YTD amount column is a measure written as =SUM(Data[Amount])
6. Try this measure

=[YTD amount]-CALCULATE([YTD Amount],PREVIOUSQUARTER(Calendar[Date]))

Hope this helps.

[Catherine84]

Hi Ashish,
There is no calendar table.. The data just comes in ytd q1, ytd q2, ytd q3 and ytd q4. But I need to calculate some Profit and loss accounts using mtd data so trying to break them up.. But not sure how to minus between the rows..

Hope u can help me ! 🙂

[Ashish_Mathur]

Hi,

Share the link from where i can download your PBI file.

[Catherine84]

Hi Ashish,

I have put up the publish the power BI in the link below:

https://app.powerbi.com/groups/me/reports/0191ed47-064f-41a9-ae40-b87f52bb2ddf?ctid=57368c21-b8cf-42...

Although my data do not have a calendar table, i tried to insert one and tried to name the Q1, Q2,Q3,Q4 into the calendar dates but still i cannot get the MTD data only. It seems to be still showing year to date data!

 

[Ashish_Mathur]

Hi,

There is no PBI file there.  Share the link from where i can download the PBI file.

[Catherine84]

Hi Ashish,
Thanks for your kind help!! I tried to publish but it will not publish with the desktop functionality.
Is there any where I can read on how to publish the pbi file?

Thanks!

[Ashish_Mathur]

Hi,

If you still need my help, you may upload the PBI file to OneDrive/Google Drive and then share the download link here.

[Catherine84]

Thanks Ashish! I will do it next time 🙂 appreciate your kind help!

[Sean · Accepted solution]

Hi @Catherine84 ,

I would suggest shaping the data better first In the Query Editor

1) select the Period column - Transform tab - click Extract - select Last Characters - type 1 - OK

2) change the Data Type of the Period column to Whole Number

3) click Group By - select Advanced

4) add these 4 groupings - RU - Account Code - Year - Period

5) for New column name - type - YTD Amount

6) for Operation - select - Sum

7) for Column - select - YTD Amount - OK

😎 Close and Apply

9) Finally create a New Column (this would be a DAX column)

DAX Amount =
VAR Company = Table1[RU]
VAR Code = Table1[Account code]
VAR CYear = Table1[Year]
VAR CQuarter = Table1[Period] // Period column is Number (1, 2, 3, 4) representing Quarters
RETURN
    [YTD Amount]
        - CALCULATE (
            SUM ( Table1[YTD Amount] ),
            FILTER (
                Table1,
                Table1[RU] = Company
                    && Table1[Account code] = Code
                    && Table1[Year] = CYear
                    && Table1[Period] = CQuarter - 1
            )
        )

DAX Amount V2 =
VAR Company = Table1[RU]
VAR Code = Table1[Account code]
VAR CYear = Table1[Year]
VAR CQuarter = Table1[Quarter] // Quarter column is Text (Q1, Q2, Q3, Q4)
RETURN
    Table1[YTD Amount]
        - CALCULATE (
            MAX ( Table1[YTD Amount] ), //Have to use MAX!
            FILTER (
                Table1,
                Table1[RU] = Company
                    && Table1[Account code] = Code
                    && Table1[Year] = CYear
                    && Table1[Quarter] < CQuarter
            )
        )

That should do it!

Now you data is shaped better and this new Amount column will show you the total per each Quarter only.

I'm sure ImkeF can probably do my last DAX step with M as well

[ImkeF · Accepted solution]

@Sean

Actually, I've written a function some time ago, that would make this task fairly easy: https://www.thebiccountant.com/2018/08/31/unravel-cumulative-totals-to-their-initial-elements-in-pow...

[Sean]

Thanks to @ImkeF we now have the M solution as well

Although she says "Referencing a previous row in DAX is still faster than my method" lets follow the steps...

Please NOTE: @ImkeF's solution does NOT require you to perform ANY of the steps in my original post!

Also you can or rather should follow her on Twitter as she posted about this topic on Aug 31, 2018

But back in Query Editor I'll continue from where I left off just to compare the DAX results to the M results.

10) Home tab - New Source - Blank Query - Advanced Editor

11) Go here and copy the function code (Ctrl+C) and Thank @ImkeF

12) Paste the function (Ctrl+V) in the Advanced Editor - Done

13) Rename the Query - fnGetPreviousRow

14) Back to Table1 - add a step after the Grouped Rows Step - click fx

15) in the formula bar enter the code provided in ImkeF's blog post here

fnGetPreviousRow(#"Grouped Rows", null, {"YTD Amount"}, {"Account code", "Year"}, null, null)

16) right-click the newly created YTD Amount.Prev column - Replace Values - null - 0 - OK

17) change Data Type to Decimal Number

18) Add Column tab - Custom Column - QTD Amount - [YTD Amount]-[YTD Amount.Prev] - OK

19) change Data Type to Decimal Number

20) you can now remove the YTD Amount.Prev column if you wish

21) Home tab - Close & Apply

As expected the results are the same!

Thanks!

I'm pretty sure we can find a DAX solution to this without Grouping By first as well. But I'm done for the day

@OwenAuger@Phil_Seamark

[ImkeF]

The pure DAX-solution could look like so: 

QtlyAmount =
Table1[ YTD Amount]
    - CALCULATE (
        MAX ( Table1[ YTD Amount] ),
        FILTER (
            Table1,
            Table1[Account code] = EARLIER ( Table1[Account code] )
                && Table1[Year] = EARLIER ( Table1[Year] )
                && Table1[RU] = EARLIER ( Table1[RU] )
                && Table1[Period] < EARLIER ( Table1[Period] )
        )
    )

This function is not as portable as the M-function, as it requires a bit more typing. 

BTW: I assumed that there is an error in the source data and there should be no duplicate attribute combinations. So replacing the last 8 rows with "ABC Co" instead of "XYZ Co". (Otherwise one should aggregate these rows in the query editor before).

[ImkeF]

Or an optimized version that calculates faster: 

QtlyAmount_Fast = 
Table1[ YTD Amount]
    - CALCULATE (
        MAX ( Table1[ YTD Amount]),
        FILTER ( Table1, Table1[Account code] = EARLIER ( Table1[Account code] ) ),
        FILTER ( Table1, Table1[Year] = EARLIER ( Table1[Year] ) ),
        FILTER ( Table1, Table1[RU] = EARLIER ( Table1[RU] ) ),
        FILTER ( Table1, Table1[Period] < EARLIER ( Table1[Period] ) )
        )

[ImkeF]

Possibly, this is even faster, please try it out:

QtlyAmount_Faster = 
VAR ThisAccountCode = Table1[Account code]
VAR ThisYear = Table1[Year]
VAR ThisRU = Table1[RU]
VAR ThisPeriod = Table1[Period]
Return
Table1[ YTD Amount]
    - CALCULATE (
        MAX ( Table1[ YTD Amount]),
        FILTER ( Table1, Table1[Account code] = ThisAccountCode ),
        FILTER ( Table1, Table1[Year] = ThisYear ),
        FILTER ( Table1, Table1[RU] = ThisRU ),
        FILTER ( Table1, Table1[Period] < ThisPeriod )
        )

[Sean]

@ImkeF

Indeed there's absolutely no need for all conditions in my original solution!

So here's the revised solution when the Quarter column is Text

DAX Amount V2 =
VAR Company = Table1[RU]
VAR Code = Table1[Account code]
VAR CYear = Table1[Year]
VAR CQuarter = Table1[Quarter] // Quarter column is Text (Q1, Q2, Q3, Q4)
RETURN
    Table1[YTD Amount]
        - CALCULATE (
            MAX ( Table1[YTD Amount] ), //Have to use MAX!
            FILTER (
                Table1,
                Table1[RU] = Company
                    && Table1[Account code] = Code
                    && Table1[Year] = CYear
                    && Table1[Quarter] < CQuarter
            )
        )

And when the Quarter column is Number

DAX Amount =
VAR Company = Table1[RU]
VAR Code = Table1[Account code]
VAR CYear = Table1[Year]
VAR CQuarter = Table1[Period] // Period column is Number (1, 2, 3, 4) representing Quarters
RETURN
    [YTD Amount]
        - CALCULATE (
            SUM ( Table1[YTD Amount] ),
            FILTER (
                Table1,
                Table1[RU] = Company
                    && Table1[Account code] = Code
                    && Table1[Year] = CYear
                    && Table1[Period] = CQuarter - 1
            )
        )

BTW I agree about the source data that there should be no duplicate attribute combinations!

That's really all for tonight folks!

[ImkeF]

@Sean s version with the quarter number has the big advantage that it caters for possible negative numbers as well!

If performance is an issue, consider splitting up the conditions into different filters like in my last code.

[Catherine84]

Thanks Sean ! It works! 🙂 really appreciate your help!

[Catherine84]

Thanks ! I will revert once I get to the computer tonight 🙂 appreciate your kind help! 🙂