Using Function Points in XP

1
Using Function Points in XP

• Andrew M. Fuqua
• Internet Security Systems
• 6303 Barfield Rd
• Atlanta, GA 30328 USA
• afuqua _at_ acm . org

2
What I wanted to know

• Would FP produce a more accurate schedule?
• Are FPs a good measure of velocity?
• Do FPs predict how long it will take to implement
  a story?

3
The answer?

• NO

4
The results

• FPs correlate poorly with estimated and actual
  task hours
• FPs/iteration not as stable as actual effort
• FPs are a poor measure of velocity
• IEDs correlate well
• a good measure of velocity
• more accurately predict project completion date

5
Why you should not believe me

• Im a novice function point practitioner
• My Logical Transactions too small
• Not a controlled experiment

6
Why were the results bad?

• Repetitive tasks inflate the FP count
• Treatment of 0 FP stories
• Usability
• Meetings
• Analysis work
• Defects
• Large refactorings

7
Suggestions

• Use FPA on an XP project

• The process helps find stories
• Forces programmers to consider I., P. and O.
• and make better estimates
• Validate the initial schedule

8
Suggestions for the researcher

• Compare XP vs non-XP team productivity
• Compare team productivity before and after
  adopting XP
• Compare team productivity on 2 code bases
• one written using XP
• the other without

9
Links

• The report
• http//groups.yahoo.com/group/extremeprogramming/f
  iles/
• XP and FPA.pdf
• United Kingdom Software Metrics Association
• MKII Function Point Analysis Counting Practices
  Manual
• http//www.uksma.co.uk