Join 34,000+ subscribers and receive articles from our blog about software quality, testing, QA and security.
 

How does the forecast works


#1

Hi,

I was wondering how exactly the forecast works.

We have a test run with around 700 tests. The forecast tells me that we are finished 08/10/2016, which is pretty late - so I was wondering how does this calculation works.

Is it simply adding the estimates and then dividing them by 8h/day or so?
Is there any possibility to include the number of people involved in the testing process ? Because the forecast of course depends on the number of testers and test cases etc…

Any inside information to share ? :slight_smile:
Thanks


#2

Hi Andreas,

The algorithm uses a mix of the Estimate and Elapsed values to calculate a forecast per test case. Given the velocity of the test run/plan (i.e., how fast results are added and tests are completed) and those forecasts on the case/test level, TestRail and the progress reports can calculate a burndown chart and projected completion date. So, the algorithm automatically adapts to changes in your testing and the projected completion date updates automatically if the velocity changes (e.g. more testers working on the same run/plan/milestone).

You can find some additional background information here:

http://docs.gurock.com/testrail-userguide/userguide-tips#scheduling_and_forecasting

I hope this helps!

Cheers,
Tobias


#3

hi,

thanks for the answer. I can see that now with more people and faster progress also the forecast changed quite to the 20/06

Seems more accurate :slight_smile:


#4

Hi Andreas,

You are welcome :slight_smile:

Cheers,
Tobias


#5

additional question.

Is the velocity you mentioned memorised over test runs ? So that, when creating a new test run, the forecast is more accurate ?

The scenario is that we finished our test run with X test cases and Y people in a timespan Z.

Now when we planning a new test run, will the system predict also Z for that ? I like to have an accurate estimate before we start the testing, to get an idea whether we need to decrease number of test cases or increase number of people - but this would only work, when the estimate for future test runs include experience from old ones

Thanks


#6

Hi Andreas,

The velocity is not remembered and this is different/calculated on a per-run/plan/milestone basis. Using past velocities is difficult as the current velocity of a run is a much better way to determine projected completion dates and past velocities are usually less meaningful. As the velocity changes, the projected completion date is updated as well and the algorithm automatically adjusts the projected completion date based on how fast your team is actually able to go through your tests. Past performance is taken into account via the forecast per test/case based on the values of the Elapsed field.

I hope this helps!

Cheers,
Tobias