How to calculate 3 week moving average using eazyBI

[Bernie Carson]

I'd like to create a plot that is updated daily with a 3 week moving average of how many story points are resolved. Mechanically this would mean looking at 3 weeks into the past and finding what the average story points resolved per day is and having that be one data point. Then the next day the same information would be pulled and a new average would be found and that would be the next data point and so forth conitnuously. If anyone has any insight on how to do this with JIRA dashboards or eazyBI it would be much appreciated. I haven't been able to find data the automates moving average graphs in connection with JIRA.

[Zane eazyBI Support · Accepted solution]

Hi,

When time hierarchy level in a report is days, then this calculation will do the work:

Avg(
  LastPeriods(21, [Time].CurrentHierarchyMember),
  [Measures].[Story Points resolved]
)

When time hierarchy level in a report is weeks, then you may use 

Avg(
  LastPeriods(3, [Time].CurrentHierarchyMember),
  [Measures].[Story Points resolved]
) 

More about dynamic time periods (including function LastPeriods() ) you may find in the eazyBI documentation.

Zane
eazyBI Customer Support Consultant

[J.M.]

@Zane eazyBI Support I want to tweak this to return me the same average but instead of story points, I'd like an issue count of resolved issues. Using the above code and swaping story points resolved for issue count only returned me the full count, and not the average. can you assist?

[Zane eazyBI Support]

@J.M., your approach to change only the measure is correct.
For example, to calculate a three-week moving average for resolved issues, you may use measure Issues resolved and put it in a formula like this:

Avg(
 LastPeriods(3, [Time].CurrentHierarchyMember),
 [Measures].[Issues resolved]
)

Make sure that in Time dimension is selected the right hierarchy over which you would like to get running average (see picture below).

Zane / support@eazybi.com

[J.M.]

Thanks! this is great! I'll be working a lot more with the tool. Is the best means on contact through email or tagging eazyBI here?

[Zane eazyBI Support]

You may use both channels. eazyBI Support is checking questions in Atlassian Community pages as well.

The community is a great power: if you ask the question there, you may get an answer from other experienced users, and this answer would benefit other users as well. Also, you may google the topic later, while, in your mailbox, it would get lost.

If you have a question specific to your business case or with sensitive information, you may reach eazyBI support via email.

[Larry Lowe]

@Zane eazyBI Support i tried the above solution and i ran into problems with reporting an accurate weekly average. Even if you look at the above image from March 13, 2018... Week 10 has 1 issue resolved, and an average of 1... Week 11 has 0 resolved, but the average is still 1... Week 12 has 0 resolved, and the average is still 1... by this point, the average should be 1/3 or 0.333.

Is this because the resolved values are blank instead of actually saying 0? Is there a formula to swap out blank values for 0?

[Zane eazyBI Support]

@Larry Lowe your guess is right. Empty values are not taken into account for average calculation. To make periods without resolved issues count as well you may use function CoalesceEmpty() in the formula like this:

Avg(
  LastPeriods(3, [Time].CurrentHierarchyMember),
  --count as zero if no resolved issues in a period
  CoalesceEmpty([Measures].[Issues resolved],0)
)

Best,
Zane / support@eazyBI.com

[larry.lowe]

@Zane eazyBI Support Thank you this works perfectly!

[Shenghan Gao]

How can I make this `Last 2 week` thing not depend on selected time dimension? I tried the following but it still gives 4 days average when the hierarchy is daily.

```

Avg(
 LastPeriods(3, [Time.Weekly].CurrentHierarchyMember),
 [Measures].[Issues resolved]
)

```

[Shenghan Gao]

@Zane eazyBI Support 

[Roberts Čāčus - eazyBI]

Hi @Shenghan Gao

Please try to use the DateMember() function to get a specific Time dimension level member from the one you have currently selected. Please have a look at the formula below:

[Time.Weekly].[Week].DateMember([Time].CurrentHierarchyMember.StartDate)

In the first part you define the Time dimension hierarchy and the level you want to use and in the DateMember() function specify the members start date from which to get the Week member. In this case the current hierarchy member.

The formula for the average calculation could look like this:

Avg(
LastPeriods(3, [Time.Weekly].[Week].DateMember([Time].CurrentHierarchyMember.StartDate)),
[Measures].[Issues resolved]
)

Roberts / support@eazybi.com

[Jeff Tillett]

When I use 

Avg(
 LastPeriods(3, [Time.Weekly].CurrentHierarchyMember),
 [Measures].[Issues resolved]
)

 I get this error:

Failed to execute query. Error message:
Internal error: member [Time.Weekly].[$total_aggregate] not found among its siblings

Any thoughts on that one? 

[Roberts Čāčus - eazyBI]

Hi @Do Not Use ,

If you are using the "Total of rows" function in the report, then please update the formula to ignore the Total row with the code below:

CASE WHEN [Time.Weekly].CurrentMember.Name <> '$total_aggregate'
THEN
  Avg(
    LastPeriods(3, [Time.Weekly].CurrentHierarchyMember),
    [Measures].[Issues resolved]
  )
END

Kind regards,

Roberts / support@eazybi.com

[Jeff Tillett]

That was it! Thank you!