Forum Discussion
Help clean up this simple DAX calculation
Hi. Given a table of values, something like below, I want a measure that will return the "within range" percentage for each category. So if 3 out of 4 values are with in range (as shown below) the measure should return 0.75 for category A.
Table1
| Category..... | RangeHigh..... | RangeLow..... | Value |
| A | 30 | 50 | 52 |
| A | 30 | 50 | 45 |
| A | 30 | 50 | 40 |
| A | 30 | 50 | 38 |
[extra dots for formatting]
What I've come up with is below, but the problems are:
- The category is hard coded -- I'd like it driven off of a slicer or filter, and
- It's messy -- I think there's a clearer more compact way to write the DAX.
PctInRange = 1 - CALCULATE(COUNTROWS(Table1), FILTER(Table1, (Table1[Value] < Table1[RangeLow] || Table1[Value] > Table1[RangeHigh]) && Table1[Category] = "A")) / CALCULATE(COUNTROWS(Table1), Table1[Category] = "A")
Can someone help? Thanks for taking a look!
A few thoughts and a question.
- In your initial COUNTROWS you'd be better just to directly COUNTROWS of the table returned from the filter.
- You can make your code a lot more readable by using variables and formatting it.
Those combined would give:
PctInRange = VAR tblGroupAInRange = FILTER( Table1, (Table1[Value] < Table1[RangeLow] || Table1[Value] > Table1[RangeHigh]) && Table1[Category] = "A") VAR cntGroupAInRange = COUNTROWS ( tblGroupAInRange ) VAR cntGroupA = CALCULATE( COUNTROWS ( Table1 ), Table1[Category] = "A" ) VAR Result = 1 - DIVIDE ( cntGroupAInRange, cntGroupA) RETURN ResultIn terms of making it more dynamic depending on how you're intending to use the measure you could just get rid of the hard coded category = A filters. Sliceing on category would pass the filter straight into both.
Lastly are the high/low values the same for all of category A? If so moving them out to a category dimension might help further.
Share a little on how you'd like to use the measure and we can take it further.
3 Replies
A few thoughts and a question.
- In your initial COUNTROWS you'd be better just to directly COUNTROWS of the table returned from the filter.
- You can make your code a lot more readable by using variables and formatting it.
Those combined would give:
PctInRange = VAR tblGroupAInRange = FILTER( Table1, (Table1[Value] < Table1[RangeLow] || Table1[Value] > Table1[RangeHigh]) && Table1[Category] = "A") VAR cntGroupAInRange = COUNTROWS ( tblGroupAInRange ) VAR cntGroupA = CALCULATE( COUNTROWS ( Table1 ), Table1[Category] = "A" ) VAR Result = 1 - DIVIDE ( cntGroupAInRange, cntGroupA) RETURN ResultIn terms of making it more dynamic depending on how you're intending to use the measure you could just get rid of the hard coded category = A filters. Sliceing on category would pass the filter straight into both.
Lastly are the high/low values the same for all of category A? If so moving them out to a category dimension might help further.
Share a little on how you'd like to use the measure and we can take it further.
I am thinking you could accomplish this using a calculated column in table 1 that tests whether a Value is in range and returns something like "isInRange" or notInRange". Then write a measure that divides the count of "isInRange" by all values in the calculated column to get the percentage. That should allow filtering and slicing by categories that are in Table 1.
I could do that but I don't want to add excessive overhead to the table if it could all be managed in DAX without impacing performance, which I think it can.
Can someone suggest how to modify the DAX, please?
