Calculating percent difference between results of two measures

[KylePerkins]

I'm attempting to create a DAX measure which tells me the % difference of results from one measure to another. The measure I've written below gives me a result, however it is an incorrect % differnce which I confirmed when I calculate the % difference of the results by hand (see screen shots below for an example and the measure titled "Measure"). I also double checked my visual filters to be sure they were the same on both visuals shown below so I believe they should be displaying the same results. Any help on why this is happening would be greatly appreciated, thanks!

Savings Percent Difference (%) =

CALCULATE('Usage v2'[Measure],'Usage v2'[Improvement Package]='Usage v2'[run type],'Usage v2'[Method]="1")

/

CALCULATE('Usage v2'[Measure],'Usage v2'[Improvement Package]='Usage v2'[run type],'Usage v2'[Method]="2")

Measure =

SUMX(

FILTER('Usage v2','Usage v2'[run type] = "base"),

'Usage v2'[Value])

-

SUMX(

FILTER('Usage v2',

'Usage v2'[Improvement Package]='Usage v2'[run type]),

'Usage v2'[Value])

The result in the graph above should be .275.

[KylePerkins]

After trying that solution this is the result 

Somehow now the result is shifted a decimal place... 

Here is the exact measure as is: 

% Savings = DIVIDE(

SUMX (

FILTER ( 'Usage v2' , 'Usage v2'[run type] = "base" ),

'Usage v2'[Value] )

-

SUMX (

FILTER ( 'Usage v2' ,

'Usage v2'[Improvement Package] = 'Usage v2'[run type] ),

'Usage v2'[Value] ),

SUMX (

FILTER ( 'Usage v2' ,

'Usage v2'[Improvement Package] = 'Usage v2'[run type] ),

'Usage v2'[Value] ))

Currently my Value column contains many different results  and I use filters specific to that visual to display results of a particular type. If I were to remove the filters the results displayed would be nonsensical. Is this structure perhaps why the results are incorrect? 

[Whitewater100]

Hi:

I think you can do this.

% Savings = DIVIDE(

SUMX(

FILTER('Usage v2','Usage v2'[run type] = "base"),

'Usage v2'[Value])

-

SUMX(

FILTER('Usage v2',

'Usage v2'[Improvement Package]='Usage v2'[run type]),

'Usage v2'[Value]), 

SUMX(

FILTER('Usage v2',

'Usage v2'[Improvement Package]='Usage v2'[run type]),

'Usage v2'[Value]))

Or in words You subtract your two measures and get result (A) 6.94 and (B) 5.44. Looks like you then subtract A-B or 6.94-5.44 to equal 1.5. Now  DIVIDE(1.5, B) or 1.5/5.44 = .275.

Again Measure A - Measure B = result

Final Measure = DIVIDE(result, Meaure B)

I hope this helps..

[KylePerkins]

After trying that solution this is the result 

Somehow now the result is shifted a decimal place... 

Here is the exact measure as is: 

% Savings = DIVIDE(

SUMX (

FILTER ( 'Usage v2' , 'Usage v2'[run type] = "base" ),

'Usage v2'[Value] )

-

SUMX (

FILTER ( 'Usage v2' ,

'Usage v2'[Improvement Package] = 'Usage v2'[run type] ),

'Usage v2'[Value] ),

SUMX (

FILTER ( 'Usage v2' ,

'Usage v2'[Improvement Package] = 'Usage v2'[run type] ),

'Usage v2'[Value] ))

Currently my Value column contains many different results  and I use filters specific to that visual to display results of a particular type. If I were to remove the filters the results displayed would be nonsensical. Is this structure perhaps why the results are incorrect? 

[Whitewater100]

Hi:

I think if you can use the separate measures in the final divide or put extra ( ) to capture the subtraction part.

Measure A = 6.94

Measure B = 5.44

Measure C = Measure A - Measure B

% Measure = DIVIDE(Measure C, Measure B,0)

or

% Savings = DIVIDE((

SUMX (

FILTER ( 'Usage v2' , 'Usage v2'[run type] = "base" ),

'Usage v2'[Value] )

-

SUMX (

FILTER ( 'Usage v2' ,

'Usage v2'[Improvement Package] = 'Usage v2'[run type] ),

'Usage v2'[Value] )),

SUMX (

FILTER ( 'Usage v2' ,

'Usage v2'[Improvement Package] = 'Usage v2'[run type] ),

'Usage v2'[Value] ))

But using the first methode where you use measure names for your two results in the DIVIDE calculation is much easier to read and debug.

If you still experience an issue, can you share any example file so I can double check? Dummy data is ok. Thank you..