The only thing I'd say to this is that Mike Cohn strongly recommends never trying to correlate actual time spent on user stories to the original story points. I don't have the Agile Estimating and Planning book in front of me, but if I recall, this advice is to avoid story points losing their goal of being measures of complexity that are independent of time. I recommend the book, it's a good read.

Kieron

I am well aware of that guideline, and I agree... HOWEVER -- my intent is to uncover gross inaccuracies in story points, not to convert story points into person-hours. Here's the scenario.

Story A is assigned 3 points.
Story B is assigned 5 points.
Story C is assigned 8 points.

If I'm using the poker planning sequence, then Story B should be approximately twice as complex as Story A, and Story C should be almost three times as complex.

Now -- if I discover at the end of the sprint that my team spent these hours completing the stories:

Story A - 3 story points, 30 hours
Story B - 5 story points, 50 hours
Story C - 8 story points, 8 hours (not a typo. 8 hours, not 80)

Then I would say that I grossly overestimated the complexity of Story C (or, equally possible, I grossly underestimated Stories A and B), and I want to know why. Did our team ignore work already done? Did we read more into the story than the customer really wanted? Do we have a genius on the team who should be cloned? Or for the underestimated tasks -- did I forget some design work? external integration? testing efforts?

If my story points don't have some level of accuracy, then I can never establish a good team velocity. One sprint we'll do 20 points, the next 50, and the next 5. Comparing the MAGNITUDE of actual hours to the MAGNITUDE of estimated points gives me a qualitative feel for the degree of accuracy in my story points.

Hi James

Have you tried writing a BIRT report? This should be able to collate the data (but have to admit I have not tried to build such a report)

anthony

Okay - I found BIRT and have it installed, and I can work through creating new reports, BUT -- it would be really helpful if I could find documentation on the data models for the "Jazz Data Source". Anyone have a source for that?

Okay - I found BIRT and have it installed, and I can work through creating new reports, BUT -- it would be really helpful if I could find documentation on the data models for the "Jazz Data Source". Anyone have a source for that?

I think this is what you're looking for: https://jazz.net/wiki/bin/view/Main/DataWarehouseSnapshotSchemas

Hope that helps.
Kieron

That's exactly what I was looking for, and I learned that the Snapshots do not expose to BIRT what I need. To actually get what I need, I would need to create new snapshots, and/or expose the "live" tables to BIRT. The twiki explains how...

But now I need access to the data schema for the live tables, not the snapshots. I browsed the wiki, but missed it if there...

Hi James,

Can you please describe to me formally what input does this report take?

And what output output should it produce?

.

Back at the start of this thread I explained my problem.

I want to query tasks that are the child of stories whose status is "Done". I can create that query now. However, in the result layout, I wish to show:

task ID
task summary
task time spent
Parent story ID
Parent Story summary
Parent Story Points

While the query builder will let me use parental values in the "where clause", it will not allow me to use any parental values in the results.