Forum Discussion
measure moving average
- 8 years ago
Hey,
here you will find my pbix file. I just added another page and created a simple table, this way was somewhat easier to read, than deduct the expected result from your graphs (wondering about the size of your screen :-) ).
I created 4 measure, two measures calculate the sum of quantities and sessions for the last 10days. The measure for quantity looks like this (session is similar)
sum quantity last 10 = var mylastdate = LASTDATE(VALUES(Calendar[Date])) var sumQuantity = CALCULATE( SUM('OrderItems'[Quantity]) ,DATESINPERIOD(Calendar[Date], mylastdate,-10,DAY) ) return sumQuantityAnd two measure that calculate the average
- the first one "avg conversion rate simple" dividing by dividing [sum quantity last 10] / [sum sessions last 10]
- the seond one "avgx conversion rate" iterates over the last 10 days and calculates the average from the divisions for each day
Here is the DAX of the first one, I recreated measures as variables to be sure to understand what's happening, I guess you can replace some of the DAX statement by simply using your existing measures:
avg conversion rate simple = var mylastdate = LASTDATE(VALUES(Calendar[Date])) var sumQuantity = CALCULATE( SUM('OrderItems'[Quantity]) ,DATESINPERIOD(Calendar[Date], mylastdate,-10,DAY) ) var sumSessions = CALCULATE( SUM('Sessions'[Sessions]) ,DATESINPERIOD(Calendar[Date], mylastdate,-10,DAY) ) return DIVIDE(sumQuantity, sumSessions, BLANK())Here is a screenshot of the table i used to validate the calculations:
One comment:
Wrong:
Using LastDate(...) -10 basically leads to a timeframe that contains 11 days, the lastdate (date of the current row) and 10 days before;-)Hopefully this is what you are looking for
Regards
Tom
Thanks for the reply.
It's not quite right. When I look at the visual, the Moving Avg doesn't line up with the underlying data. ie- a big spike in Conversion Rate doesn't reflect a big spike in MA 10 days later.
ALSO- when I do the math manually (ie- average the Conversion Rate over a 10 day period), the MA number doesn't match the number I am getting.
Your solution SEEMS like it would work, but, not when I apply it. Any ideas?
Hey,
sorry to hear, that it looks good but behaves not good enough :-)
Maybe you might to consider to provide a pbix with sample data, upload the file to onedrive or dropbox and share the link
Regards
Tom
- PBINoob8 years agoHelper I
Here's a quick image of what I'm seeing-
If you average the conversion rate from 4/1 to 4/10, you get 34.67% (including 4/1 to 4/10). BUT- on 4/10 the Moving Average using your formula is 35.0%. If it's using the PREVIOUS 10 days- that math doesn't work either?
ALSO- if you look at the highest spike on the bar graph- 10 days later is a pretty low dip in the line (line is the MA).
What do you think?- TomMartens8 years agoSuper User
Hey,
please provide a link to a PBIX with sample data
Regards
Tom
- PBINoob8 years agoHelper I
Sorry- how do I do that? Make a copy of my file, and then delete out all the tables, relationships, etc- and manually fill in the data with false data?