| || |
Figures on Capers Jones equation
This is a question for those who use FP as an estimation measure:
Have you used Capers Jones equation on a project ? If yes, I'd like to hear a comment on the accuracy of the numbers you got vs the actuals.
Re: Figures on Capers Jones equation
I have used it and variations of it. Personally, I found the best accuracy (if going by function points - which is not something I always like to do) by figuring out the function points by the familiar process of estimating the number of external inputs, external interface files, external outputs, external queries, and logical internal tables. Usually those are a raw (dimensionless) value for function points and thuse are not "actual" function points. So you multiply your base number by a given conversion factor. (Those are usually based on Capers Jonses' work.) I still find you can run into trouble with that, however.
So one thing that I like is the "backfire" approach where you convert the function points to SLOC or ELOC. The problem is that you need a table of language equivalencies to do this. (Of course, you could derive your own, particularly if you are using a lot of third-part components.) From this you can easily see how to work out productivity and effort values (such as in person-months).
You can also use things like class-method points for object-oriented languages or use case points, if those are being used, although I find use case points can become very vague if the system is reasonably complex.