Issue history prediction chart (3 worriers)

Questions & Answers
1

Bonjour tout le monde,

I’m making a graph to display the history of tickets to do, in progress and finalized. On the graph, I’d also like to display the day’s line, then draw prediction curves based on figures from previous weeks.

However, I have 3 worries:

  • issues history curves continue after the current date. Is there a trick or do I have to create a new measure with a case when to check the date?
  • For the vertical today line on the chart it’s ok. Except that I actually have 3 vertical today lines. It’s not very clean in the legend. Is it possible to hide some legends?
  • More complex (perhaps). My prediction doesn’t really reflect the prediction based on the latest trends. The last few weeks have seen the number of “to do” tickets drop: 95-89-81-86-70. I wonder why the prediction doesn’t go down. Probably because of my formula :sweat_smile:

Here is my prediction formula

CASE WHEN
DateCompare(now(),[Time].CurrentHierarchyMember.StartDate) <= 0
THEN
LinRegPoint(
DateToTimestamp([Time.Weekly].CurrentMember.StartDate),
LastPeriods(5, [Time.Weekly].CurrentMember),
[Measures].[Issues history],
DateToTimestamp([Time.Weekly].CurrentMember.StartDate)
)
WHEN
DateInPeriod(‘Today’,[Time].CurrentHierarchyMember)
THEN
NonZero([Measures].[Issues history])
END
END

Thanks you very mutch, I’m interested in any tips, I know a bit about EazyBI but there’s still so much to learn :slight_smile:

image
image972×645 11.9 KB

image
image1561×365 37.5 KB

2

Hello @jerome_lgr,
Great job on your report!

About your questions:

  1. Yes, you should create a new measure that checks the date and shows you “Issues history” only if it is before today. The formula would be like this:
CASE WHEN
  DateAfterPeriodEnd(
    Now(),
    [Time].CurrentHierarchyMember.PrevMember
  )
THEN
  [Measures].[Issues history]
END

  1. You need to use the option “Drill into measure by another dimension level” for the “Issues history” and “Issues history prediction” measures. The “Today” won’t be split by the “Transition Status” dimension.

  2. In the formula, you are checking only the last five weeks; thus, the prediction with each future line will be more similar to the issues history measure (the same number for the future periods). Maybe it is worth checking those prediction measures from this eazyBI Demo report: Project Prediction report - Issues - Jira Demo - eazyBI

best,
Gerda // support@eazybi.com