Thursday, May 31, 2007

Software Estimation - Lines of Code (Part 1)

One of the primitive and efficient method for estimation of software development time is based on the Source Lines of Code (SLOC). For maintenance or enhancement projects the SLOC is readily available and the impact of the new change request can be had. On the other hand, this luxury is not available for new projects - projects from scratch. Here we need to assume to SLOC and start working

Now the question is 'how to get the Source Lines of Code more accurate?'

We need to get the figures from the experts who know the code and will be doing the coding - the programmers, the DBAs etc. It is advisable to get the figures from them for three different situations

  1. Expected
  2. Best
  3. Worst
Expected will be the usual case and the Best scenario will less estimated time and the worst, obviously have more time .. if a module gets screwed up etc..

The final figure of SLOC can be achieved by using the Mean and Standard Deviation

Mean =[best+worst+4*(expected)]/6

and the Standard Deviation can be achieved by

SD
=(worst–best)/6

Final Estimation = Mean + 3 * SD

The final estimation will give a bit accurate figure.

The estimation of lines of code should not count the following:

  • Comments
  • Blank lines
  • OS include files directives etc

  • There are many tools to do the estimation. We will discuss on some of them in the future articles.

    For mean and SD, Microsoft Excel should come handy. You can also write some custom functions in MS Excel (http://vbadud.blogspot.com) for customized calculations







    Digg Technorati Delicious StumbleUpon Reddit BlinkList Furl Mixx Facebook Google Bookmark Yahoo
    ma.gnolia squidoo newsvine live netscape tailrank mister-wong blogmarks slashdot spurl