H2O: Generalized Linear Model
The H2O airline dataset over on GitHub offer a great intro to leveraging GLM to make predictions. Which leads to the following possibly data sets which could possibly have predictions made on them:
- Project/team/iteration – successful iteration delivery, number of defects post iteration done
Essentially the above is driven by the Skills Cloud postings I made some time ago. If you have an organisation modelled from the perspective of projects and teams, can you now begin to make an assessment around team productivity? Part of the underlying issue around these data sets, unlike historical airline data sets, is that there is possibly a higher chance of the data being gamed.
Taking the project/team/iteration outcome, it would be interesting to see if using various algorithms/models, it was possible to predict with any degree of certainty if a team would deliver on an iteration, or if a team member being sick, impacted the team itself from delivering successful “Done” stories.