Velocity Learner

How many points should we sign up
for in the next sprint ?

Imagine you are responsible for scrum or iteration planning. You might be a scrum master. You use story points, or functions points, or some other measure such as lines of codes to determine the team velocity: that is, how much code can be developed in a particular scrum or iteration. A typical question you face is:

What team velocity should I use for planning?

You know that your team has completed three sprints with the following results:


Table 1: Historic Velocity

Sprint Story Points
1 36
2 42
3 58
4 34

The simplest and most common way to answer the question is to determine the average velocity—in this case, 42.5 story points per sprint. This is not a certain value, however; it is only the most likely value, sometimes called the expected value. Using this value in your planning means you will miss your deadlines about half the time.

The Aptage Velocity Learner uses modern machine learning techniques to provide you with the odds of success over a range of velocities: the lower the velocity, the higher the odds. (Mathematically curious individuals, click here.)

Figure 1 below, is a screen shot of the Aptage velocity webpage, where we have submitted the example velocities shown in Table 1. The app returns the confidence of success for each choice of velocity used in the plan. Note that, as expected, there is a 50% confidence at the average 42.5.  

Figure 1: View of the Aptage Velocity Learner Tool

The chart in Figure 2 is color coded as follows:

  • Blue: Any velocity leading to a greater than 75% likelihood of success. These are safe velocities, with only occasional slips.
  • Green: A velocity with likelihood of success between 50% and 75%. These are reasonable velocities that will keep the team productive but with more-frequent slips.
  • Yellow: Velocities with confidence between 50% and 25%. These can be considered aggressive and most likely will lead to slippage.
  • Red: Velocities with confidence less than 25%. These are simply a bad bet and should be avoided.

In our example, the safe bet is 32, the reasonable bet is 43, and anything above 54 is a high risk.


Figure 2: Estimated Velocity Chart

What to test our beta version?