This project:

Project Link: Bayes Estimation

Provides a set of notes and an example implementation in R for performing linear regression and using bayes rules to update the parameter distributions as well as some discussion of the method.

It is a briefer summary of some of the study I was doing that year on the topic.

The domain that it is applied to is the heady topic of software project estimation or rather, guesstimation.

I would not take the software duration estimation part too seriously, as it was a vehicle for the discussion in order to demonstrate the method rather than the actual focus.

These days it seems the #noestimates movement has pretty much made the subject of project estimates in software engineering moot. Although it does not necessarily mean you should not attempt software project estimation, rather, a frank discussion about the limitations of project estimation in the domain of software engineering is possibly worthwhile having prior to an attempt. However, if you do have to do it, it might be worthwhile considering a statistical approach ;) Although the challenges of collecting enough data and valid attributes are only summarily idenfitied in the discussion linked to above.

There are also some interesting studies using linear methods for software engineering related topics, one area that demonstrated the application in a very practical way was on the topic of code quality and leveraged generalised linear methods for their purpose. This paper, published in the CACM edition for October, was quite a good discussion of the general methodology they used and interpretation of their results. And is an interesting example of a statistical analysis being applied to the question of software engineering quality.