Skip to content
Radek Falhar edited this page Jan 10, 2020 · 7 revisions

Jira Flow Metrics

Jira flow metrics is a WPF .NET application designed to query issues from Jira through API and then analyze them using flow metrics like Cycle time, Work in Progress in charts like Cumulative Flow Diagram, Cycle Time Scatterplot and Cycle Time Histogram.

Usage

Download latest release from Releases.

The application requires .NET 4.7.2 installed to work correctly.

The release a simple ZIP file with binaries. There is no installer yet. Simply extract the files anywhere. Start the application using JiraFlowMetrics.exe

Jira update

First screen is to connecting Jira, downloading issues, and caching them locally. To update issues from Jira, enter Jira's URL, user name and password and key of project to load issues from. First time update is run, all issues from given project are loaded into local cache. Subsequent updates only load updates for issues that were changes since last update. Updating first time might take few minutes. Subsequent updates should be faster.

The local cache exist primarily as performance optimization, as querying necessary information for relevant issues can take lots of time.

Only currently supports single project. Updating for multiple projects will create corrupted data.

Update screen

Story filtering

Right now, there is hard-coded filter where only specific issues are used. These are stories of Type Story or Bug, which's Resolution is not Cancelled or Duplicate and which's Status Withdrawn or On Hold. Issues that fit this filter are marked as valid.

Story state and filtering

One difficulty of analyzing state transitions and cycle time is that states can go back in Jira. This is not allowed for any of the subsequent analysis. This screen contains settings to fix this kind of behavior. Simply use arrows to setup appropriate order of states for "Filtered states" from available states. When an issue's state is analyzed, the first time issue's state is changed to given state, it stays there and doesn't change even when issue's original state goes back in the list. The issue is considered complete when last state is reached. Cycle time is calculate as time between first and last state.

Another difficulty is that stories can be set to states that don't indicate any work or activity. Reset states is list of such states. Once issue is set to this state, all of it's progress is reset and starts again. This can happen multiple times.

Story filtering screen

Issue grid

Display all issues and their values in grid. Started, Ended and Duration are start, end and length of cycle time.

Issues grid screen

Cumulative flow diagram

Displays cumulative flow diagram of all filtered states.

Date range is currently hard-coded for 3 last months.

Cumulative flow diagram screen

Cycle time scatterplot

Scatterplot of cycle times of all valid and completed issues. Blue are stories, Red are bugs.

50%, 70%, 85% and 95% are calculated and displayed. Either as horizontal lines or as labels bellow the chart. Due to bug in charting library, it is not possible to display the values next to the horizontal lines themselves.

Cycle time scatterplot screen

Cycle time histogram

Histogram of cycle times of all valid and completed issues.

Cycle time histogram screen

Cycle time and throughput history

Display history of how cycle time and throughput changed through time. Each day represents cycle times (50%, 70%, 85%, 95%) of a past period. Throughput is how many issues were finished in that same period. Right now, the displayed period is a year and the calculation window is 3 months.

Cycle time and throughput history

Story point vs. cycle time

Displays relations between cycle times and estimated story points. Cycle times are grouped according to story points and displayed with count, average cycle time, 5%, 25%, 75% and 95% of cycle times.

Story point vs. cycle time screen

Simulation

A monte-carlo simulation based on cycle times and issue creation rate. When run, shows histogram of how long would it take to finish given count of stories.

Simulation screen