Conversation
* Added a cycle time histogram tracking frequency of cycle times
* Cycle times rounded to nearest 10th of "day" (flooring hundredths)
bin size
* Bin sizes automatically calculated (chart split into 25 bins) * 50th percentile marker on histogram
|
I've now made the bin sizing a little more intelligent (once there are more than 25 samples it starts using 25 bins, the size of the bins gets dynamically updated as the number of cycle times increases). I've also added a 50th percentile marker on the cycle time histogram - this can be slightly different to the average shown on the main stats as its set at the point where 50% of cycletimes are to the left of the line so it always aligns with a cycletime in the data whereas the average may not. |
* 85th percentile marker as an example for service level expectation
for team
|
Screenshot of new binsizes, 50% & 85% markers. The two percentiles are useful for talking about the "coin toss" probability of delivering within the average cycle time & the 85% (as an example) that could be used for a team to publish some sort of service level expectation (SLE).
|
1 participant
This adds a cycle time frequency histogram to the graphs plotted as it was useful to show the distribution of cycletimes when explaining things to teams (especially when starting to talk about SLEs etc, expect another pull request with percentile markers etc)
At the moment it buckets cycletimes into tenths of "days" (bin size) so the histogram is a little more clear. You could do this more intelligently, but this works for now.