Calculation Between Two Columns - Rolling 24 Months - New Month Gets Added, Oldest Month Drops Off

[bvbull200]

I've hit a roadblock on a report I'm working on.

I have data like the attached.  It shows a list of locations, then has the revenue for each of the past 24 months and the number of business days in each of the last 24 months.  With this, I can calculate daily average growth (DAG). 

With static data, it is easy to do the following calculation:

(((Revenue AUG 2019)/(Business Days in AUG 2019))-((Revenue AUG 2018)/(Business Days in AUG 2018)))/(Revenue AUG 2019)/(Business Days in AUG 2019)

The problem is, the oldest month on the report will drop off and be replaced by the most recent month.

In this case, it means that, next month, SEP 2017 will fall off and be replaced by SEP 2019.  As such, I can't build a calculation off of existing header names - they will be changing month to month.

Is there a way to do a calculation based off of the column number rather than the column name?  For instance, take the above calculation and replace header names with a column number?

Is there another solution I should be searching?

Ideally, when I am done, I will have Revenue, Business Days, and DAG for every sales location for the last 12 months.

This is a sample of the data.  It goes out to 48 columns in the query.

Location	Revenue AUG 2019	Business Days AUG 2019	Revenue JUL 2019	Business Days JUL 2019	Revenue JUN 2019	Business Days JUN 2019	Revenue MAY 2019	Business Days MAY 2019	Revenue APR 2019	Business Days APR 2019
Alabama	1382689.5	21	1269324.34	21	1369134.09	21	1078965.78	21	1000799.88	21
Alaska	1458752.43	21	1409564.96	21	1272245.54	21	1347723.84	21	1127298.6	21
California	1620720.21	21	1334704.87	21	1040614.8	21	1201193.03	21	1182246.59	21
Deleware	1978296.83	21	2291016.27	21	2109485.17	21	2190225.58	21	2106644.09	21
Florida	968349.04	21	1119913.61	21	979007.91	21	1078147.98	21	1199920.92	21
Georgia	1741838.23	21	1869439.93	21	1605837.01	21	1854699.02	21	1903840.14	21

[Icey · Accepted solution]

Hi @bvbull200 ,

You can transform your data structure to get what you want. I just choose two months data to test. This is my PBIX file.

1. Delete ‘Changed Type’ step after ‘Promoted Headers.’

2. Select column ‘Location’, unpivot other columns.

3. Split column ‘Attribute’.

4. Change type.

5. Pivot column ‘Attribute.1’

6. Rename column.

Then you can create columns and measures.

Columns:

Year = YEAR('Rolling 2 Months'[Date])

Month = MONTH('Rolling 2 Months'[Date])

Measures:

Measure1 =
DIVIDE (
    SUM ( 'Rolling 2 Months'[Revenue] ),
    SUM ( 'Rolling 2 Months'[Business Days] )
)

Measure2 =
VAR previous_year =
    MAX ( 'Rolling 2 Months'[Year] ) - 1
VAR month =
    MAX ( 'Rolling 2 Months'[Month] )
RETURN
    CALCULATE (
        MAXX ( 'Rolling 2 Months', [Measure1] ),
        FILTER (
            ALLEXCEPT ( 'Rolling 2 Months', 'Rolling 2 Months'[Location] ),
            previous_year = 'Rolling 2 Months'[Year]
                && month = 'Rolling 2 Months'[Month]
        )
    )

DAG = DIVIDE( [Measure1]-[Measure2] ,[Measure1])

Then I just change the columns name and refresh in Power BI Desktop.

Best Regards,

Icey

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

[bvbull200]

I'm having a problem with the DAG function.

When I run the most recent month (1 month only), it works fine.  As soon as I expand in to more than 1 month, the data gets way off.

In the table you provided, the sum total of measure 1 and measure 2 doesn't make sense to me and I think whatever is happening there is causing the problem.

When I look at my Measure1 right now, every value is 50,000 or higher (like in the example above) and there are 42 rows, but the table sum at the bottom is just 58,905.88.  The revenue column appears to be adding up correctly, just not the measures.

Any idea what is going on there?

The data for a given sales office for measure 1 and measure 2 appears to be correct.  When I sum up Measure1 manually, then sum up Measure2 manually and do the math of - (Measure1 - Measure2) / Measure2 - manually, I get the right answer.  The measure being done through the PowerBI measures are yielding a different result, however.

[bvbull200]

Many thanks.  This has done exactly what I needed it to do.  I appreciate the insightful reply.  I will mark as an accepted solution!

[bvbull200]

This looks great!  I am going to run through it here shortly and see if I can make this workable.

I appreciate the detailed response.  I will report back once implemented.