Forum Discussion
Calculate difference using a ranking column
Hello all. I have some cumulative sales data which publishes every week, and I need to figure out the delta in such a manner I can easily target the larger books/debooks. Lets say my data look like this:
| Group | SubGroup | Continent | Country | Sales | Week ending | Order |
| Fruit | Apples | Africa | Zimbabwe | 100 | 21/07/2023 | 3 |
| Fruit | Apples | Asia | Fiji | 120 | 14/07/2023 | 2 |
| Fruit | Oranges | Asia | Vietnam | 300 | 7/07/2023 | 1 |
| Fruit | Apples | Asia | Fiji | 250 | 21/07/2023 | 3 |
| Veg | Carrots | Australasia | New Zealand | 10 | 14/07/2023 | 2 |
| Veg | Carrots | Europe | Czech Republic | 150 | 7/07/2023 | 1 |
Subgroups are always subsets of groups, and countries are always subsets of continents. The week ending field is text and tends to get a bit freeform, so I'm generating the 'Order' column where oldest is smallest and the week ending:order relationship is 1:1. The sales number is cumulative of all prior weeks/'orders' in the dataset.
In this example we can see that in the last week we've sold 130 fijian apples, and it's this number I wish to calculate to display in a matrix that the user can drill into .
I previously asked a similar question and tried to modify that to suit my new question, but I'm missing a fundamental concept as while my totals are correct, the values at a continent/country level are preposterously large
Thanks 🙂
2 Replies
unusdusrname , A column that can give you week sales from cumulative
sales this week=
var _date = maxx(filter( Table,[Group] = earlier([Group]) && [SubGroup] = earlier([SubGroup]) && [Continent] = earlier([Continent])
&& [Country] = earlier(Country) && [Week ending Order] < earlier([Week ending Order])),[Week ending Order])
return
[Sales] - Sumx(filter( Table,[Group] = earlier([Group]) && [SubGroup] = earlier([SubGroup]) && [Continent] = earlier([Continent])
&& [Country] = earlier(Country) && [Week ending Order] =_max ),[sales])Measure to compare this week vs last week
Week Rank = RANKX('Date','Date'[Week Start date],,ASC,Dense)
OR
Week Rank = RANKX('Date','Date'[Year Week],,ASC,Dense) //YYYYWW format
These measures can help
This Week = CALCULATE(sum('Table'[Qty]), FILTER(ALL('Date'),'Date'[Week Rank]=max('Date'[Week Rank])))
Last Week = CALCULATE(sum('Table'[Qty]), FILTER(ALL('Date'),'Date'[Week Rank]=max('Date'[Week Rank])-1))to get the week start/week end of Friday, refer
2.Any Weekday Week - Start From Any day of Week
https://community.powerbi.com/t5/Community-Blog/Any-Weekday-Week-Decoding-Date-and-Calendar-2-5-Power-BI-Turning/ba-p/1187482
https://medium.com/chandakamit/cheat-sheet-any-weekdays-week-start-date-just-one-variable-apart-6b2e6f593958Thanks Amit. I didn't quite get your code to work however, PowerBi reports syntax errors.
So I expunged the surplus carriage returns, and added the square braces around [Country]. I also changed [Week Ending Order] for just [Order] as those two were different fields.
I've still got a syntax error in the return clause where [Order] = _max - am I right in thinking _max is intended to be a variable that's been defined? So it maybe ought to be _date defined in the var clause above? Here's what I tried.
sales this week = var _date = maxx(filter( Table,[Group] = earlier([Group]) && [SubGroup] = earlier([SubGroup]) && [Continent] = earlier([Continent]) && [Country] = earlier([Country]) && [Order] < earlier([Order])),[Order]) return [Sales] - Sumx(filter( Table,[Group] = earlier([Group]) && [SubGroup] = earlier([SubGroup]) && [Continent] = earlier([Continent])&& [Country] = earlier([Country]) && [Order] =_date ),[sales])Just gotta love how PowerBI won't share even which line holds the supposed syntax error. Can you see what is wrong?
