Forum Discussion
Difference between 2 bars
Hello,
I need help in creating something similar (kind of waterfall model) visual in PBI. where we want to showcase the difference between 2 bars like this. Data contains more than 1000 rows so I have placed it here as Sample Data.
Sample data looks like this
| Cost Center Level ID | Versions | Time Period | Value |
| 10004000 | FC3 | Jan-2022 | 6,226 |
| 10004000 | FC3 | Feb-2022 | 5,685 |
| 10004000 | FC3 | Mar-2022 | 714,108 |
| 10004000 | FC3 | Apr-2022 | 9,382 |
| 10004000 | FC3 | May-2022 | 11,524 |
| 10004000 | FC3 | Jun-2022 | 689,278 |
itsmeanuj My guy. You just take the Diff measure and break it into 3 measures, one for each difference value. PBIX is attached below signature.
IP Diff = VAR __Axis = MAX('Versions'[Version]) VAR __Result = SWITCH(__Axis, "IP Diff", VAR __1 = SUMX(FILTER('Table',[Versions] = "IP"),[Value]) VAR __2 = SUMX(FILTER('Table',[Versions] = "FC1"),[Value]) VAR __Result = IF(__1 < __2, __2 - __1, __1 - __2) RETURN __Result, BLANK() ) RETURN __Result FC1 Diff = VAR __Axis = MAX('Versions'[Version]) VAR __Result = SWITCH(__Axis, "FC1 Diff", VAR __1 = SUMX(FILTER('Table',[Versions] = "FC1"),[Value]) VAR __2 = SUMX(FILTER('Table',[Versions] = "FC2"),[Value]) VAR __Result = IF(__1 < __2, __2 - __1, __1 - __2) RETURN __Result, BLANK() ) RETURN __Result FC2 Diff = VAR __Axis = MAX('Versions'[Version]) VAR __Result = SWITCH(__Axis, "FC2 Diff", VAR __1 = SUMX(FILTER('Table',[Versions] = "FC2"),[Value]) VAR __2 = SUMX(FILTER('Table',[Versions] = "FC3"),[Value]) VAR __Result = IF(__1 < __2, __2 - __1, __1 - __2) RETURN __Result, BLANK() ) RETURN __Result
15 Replies
- Greg_DecklerCommunity Champion
itsmeanuj If the standard waterfall chart won't work you'll need to check out the App Store for custom visuals that may work or use something like Deneb custom visual.
- itsmeanujHelper IV
Greg_Deckler - So is it not possible to play around with some measures to show data in this visual format? like the difference between 2 bars?
- Greg_DecklerCommunity Champion
itsmeanuj Oh, it might be possible given a disconnected table, a stacked bar chart and a fancy measure but it's not going to be easy. I'll see if I can pull something out of the hat.