4.7 TECHNIQUES FOR ESTIMATION

 < Day Day Up > 



4.7 TECHNIQUES FOR ESTIMATION

Having discussed the model-based approach to cost estimation we should now consider a different and, I believe, complementary approach. I call this approach Technique Based and think of it as "People-Oriented." You will see, when I discuss specific techniques in the context of the estimation process, that some elements of the model based approach also creeps in. The difference is that here we are really considering estimation as a people driven activity, if you want as a slightly technical art, rather than as a formal, repeatable, scientifically based activity.

The points to note about the people-oriented approach is that they take more time than simply cranking parameters through a model and that the techniques used often appear to be very simplistic, almost trivial. The most important point to note is that they also work.

The people-oriented approach and the techniques it encompasses have two main aims. First, they make estimators think about what they are doing and introduce some degree of discipline to the estimation activity. Second, they encourage communication by encouraging estimators to seek assistance from other people rather than leaving them to estimate in isolation.

Which is all I have to say about this approach to estimation just now. As you will see, it is this approach that forms the backbone of the estimation process template which is where all this talk is leading us.

Now many pundits claim that it can take ten, fifteen or twenty years for an organization to develop such models because it takes that long to get sufficient data. Experience within a few organizations suggests that this is not so. One retail organization, JC Penney, IFPUG(1) , in the US took three years to develop a cost estimation model, to validate that model and to go public with the statement, supported by statistical data, that it gave them as accurate a set of predictions as those claimed by the leading generalized cost models.

Another organization, again in the US, claimed in passing at another IFPUG conference that their local cost estimation model which included factors for requirements growth, gave them an accuracy of plus or minus five percent, that's +/- 5%, against actuals for predictions made at the end of requirements specification. It only took them five years to get this good!

Now what was common to those models? They were local, they both used Function Point Analysis as the size parameter and they both had a limited set of cost drivers. The other thing that they share is that they worked! Oh, they also took a lot less than ten years to develop! The other thing that these two local models shared is that they both fitted into an estimation strategy that went beyond the models.



 < Day Day Up > 



Software Metrics. Best Practices for Successful It Management
Software Metrics: Best Practices for Successful IT Management
ISBN: 1931332266
EAN: 2147483647
Year: 2003
Pages: 151
Authors: Paul Goodman

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net