7

One of the main questions about Story Points is what is their accuracy. Has any research or survey been performed to analyse their precision?

For example what is the accuracy of the completion date estimation based on Story Points.

Chris Brettini
  • 970
  • 9
  • 19
  • 1
    To answer this question we just need to say yes (giving at least one example) or no (explaining how we searched)? – Tiago Martins Peres Jan 26 '20 at 10:44
  • Links to researches would be appreciated!) – Chris Brettini Jan 26 '20 at 10:51
  • By 'precision', do you mean 'accuracy'? Because I can answer your Question as-is with two words - 'Fibonacci, usually'. – Sarov Jan 27 '20 at 14:12
  • @Sarov Yes, I mean "accuracy" – Chris Brettini Jan 27 '20 at 19:36
  • 2
    Interesting question. I'd be willing to bet that the cost of calibration vs the return on calibration is better for story points than for estimation by hours. I'm also willing to bet that even poor quality story point estimation by the team is superior to estimates developed by the PM in isolation. I think your question is good - but I think it has to be in the context of some questions that are much more important. – MCW Jan 28 '20 at 16:51
  • I don't think completion date and story points are related, or at least they shouldn't be related. Maybe I'm wrong, but I see SP as an estimation of effort and complexity, and not time, while completion date is purely time-related. – Daniel Jan 28 '20 at 19:30
  • 1
    I've created simple application that reads JIRA tickets, calculates their cycle time and compares it to estimated story points. When I run it on my current projects, I find minimal relationship between the two. I would bet majority of projects using story points would be same. You can try it here : https://github.com/Euphoric/JiraFlowCharts/wiki – Euphoric Jan 29 '20 at 14:04
  • @Euphoric You'd advise you to write a complete answer and explain your research in more detail. Do you include idle-time in cycle time? Inorder to calculate a time spent on an issue you would need to track time, not just to measure the start time and the end time. – Chris Brettini Jan 30 '20 at 20:21

2 Answers2

8

I know of at least two

Tiago Martins Peres
  • 2,084
  • 2
  • 14
  • 32
3

Effort Estimation in Agile Software Development using StoryPoints gives a good overview of the story points accuracy: The cone of uncertainty is inherent to estimation.

For improving the estimation accuracy, various optimization techniques have been proposed. The Support Vector Regression (SVR) is one of these techniques that helps in getting optimal estimated values and the following article explains the theory in a thorough manner.

https://ksiresearchorg.ipage.com/seke/seke14paper/seke14paper_150.pdf

Also, to improve the accuracy of effort estimation, this article proposes a model for agile software development project prediction using Bayesian networks.

Mike Cohn has invested a lot of time researching the subject of Story points. You may also look into the following articles:

Zoe Marmara
  • 348
  • 2
  • 7