Page MenuHomePhabricator

Define a TPG library of diagnostic charts/metrics
Closed, ResolvedPublic

Description

  • bug trends
  • theoretical # of bugs to be found
  • lifespan of triage tasks
  • age of submitted patches

Event Timeline

JAufrecht claimed this task.
JAufrecht raised the priority of this task from to Needs Triage.
JAufrecht updated the task description. (Show Details)
JAufrecht added a project: Team-Practices.
JAufrecht added subscribers: JAufrecht, Aklapper.
ggellerman set Security to None.

In the [[ https://docs.google.com/spreadsheets/d/1oy83WsZLFFlBY4HIazXZrVv1UKrCWfoPKCoFii9LIYw/edit#gid=1615990930 | Process Menu Google Doc ]]I created mockups of

  1. Burndown
  2. the broken Burndown currently in Phab (where committed changes retroactively)
  3. Sprint Burnup
  4. Sprint Burnup with Cumulative Flow
  5. Product Burnup
  6. Product Burnup with estimation uncertainty
  7. Velocity over time

I found more examples here, which closely overlapped everything I had in mind but had some more details and variants. I would still like to do:

  1. Cycle History (scatter or line plot of average cycle time per sprint)
  2. Cycle Spectral Analysis aka Histogram (# of stories vs cycle time)
  3. Cycle Variance (as history, but minus the mean)
  4. Story size vs cycle time
  5. Bug History (# New, triaged, done over time)

I added three of the charts to the TPG Glossary
I added a few paragraphs introducing charting the TPG handbook.

I'm ready to call this done for now; it's not 100% complete, but it's plenty to have in our virtual library for now, until we can make more of these charts for real and see what's actually useful for our teams.

  • bug trends
  • lifespan of triage tasks
  • age of submitted patches

I'd like to understand what exactly those items mean, in which systems (Phab, Gerrit, ...), so this task does not overlap with work being done in T94578.
Or is this task only about defining such metrics and not about actually implementing them?

Yes, this task should refer only to defining the charts and underlying metrics. I'll clarify the title.

JAufrecht renamed this task from Make (or find) more diagnostic charts to Define a TPG library of diagnostic charts/metrics.May 21 2015, 6:30 PM

Looks good!

  1. Product Burnup

Could you add a legend for the red and green trend lines in the Prod Burnup to the right of the VE burnup (in Glossary entry as well?)

  1. Would you open a new task to cover these?

Cycle History (scatter or line plot of average cycle time per sprint)
Cycle Spectral Analysis aka Histogram (# of stories vs cycle time)
Cycle Variance (as history, but minus the mean)
Story size vs cycle time
Bug History (# New, triaged, done over time)

FYI: Analytics Eng discussed writing a scrip to track cycle time on their Kanban board

  1. Excellent glossary entries! Only comment would be to mention the sprint commitment in the Burndown Chart section and also the good practice of closing tasks/stories as they are completed to make the slope of the line more realistic and the chart better able to convey the likelihood of making the commitment in a given iteration.

Reviewed by Grace. New tasks will cover new work. Closing