Automated Software Cost Estimation

1
Automated Software Cost Estimation

• By James Roberts
• EEL 6883
• Spring 2007

2
Background

• Over 53 of software projects overrun by more
  than 50 in both budget and schedule
• Software overrun in budget is a failure
• Software overrun in schedule is a failure
• Goal of software engineering is to deliver
  software on time and within budget

3
Possible Solution

• Automated Software Cost Estimation
• Look at history
• Generalize data
• Create equations
• Parametric

4
Input Measurements

• SLOC Source Lines of Code
• DSI Delivered Source Instructions
• Function Points

5
Cost Estimation Models

• COCOMO 81
• COCOMO II
• REVIC
• SLIM
• Others

6
COCOMO

• Developed by Barry Boehm in 81
• Based on historical database
• DSI is the input
• Three versions
• Basic Model
• Intermediate Model
• Detailed Model

7
COCOMO II

• Updated the COCOMO 81 model
• Allows for
• Spiral development
• Rapid prototyping
• COTS integration
• OO Design
• Uses SLOC

8
REVIC

• Revised Intermediate COCOMO
• Developed by Ray Kile
• Updated to use Air Force project data
• Adds a mode for Ada development
• Inputs are the same as COCOMO 81

9
SLIM

• Software Life-Cycle Model
• Developed by Larry Putnam
• Uses a Rayleigh distribution
• Project personnel vs. Time
• Intended for large projects
• Fewer parameters

10
QSMs SLIM Tool

• Based on the SLIM model
• Windows based
• Easy to use
• Several different wizards for quickly generating
  an estimate
• Five steps to create an estimate

11
Softstars CoStar

• Based on the COCOMO model
• Windows based
• Easy to use
• Many different COCOMO variations
• Create Estimate Wizard
• Many parameters required
• Highly configurable
• Full featured demo version available

12
Galoraths SEER-SEM

• Based on proprietary COCOMO-like models
• Windows based
• Moderately easy to use
• Create Estimate Wizard
• Few parameters required up front
• Highly configurable
• Poor demo version

13
Conclusion

• Would recommend the Softstar CoStar software
• Software Cost Estimation is important for any
  program manager
• These tools are vital to quickly generating
  estimates for success

14
References

• 1. Dave Srulowitz, M.B., Vic Helbling. Software
  Estimation. 2001 cited Available from
  http//www.saspin.org/SASPIN_Apr2001_COCOMO.pdf.
• 2. Briand, L.C., et al. An assessment and
  comparison of common software cost estimation
  modeling techniques. 1999.
• 3. Boehm, B.W., Software Engineering Economics.
  1st ed. 1981 Prentice-Hall.
• 4. COCOMO II. cited Available from
  http//en.wikipedia.org/wiki/COCOMO_II.
• 5. Boehm, B.C., B. Horowitz, E. Madachy, R.
  Shelby, R. Westland, C. An Overview of the
  COCOMO 2.0 Software Cost Model. in Software
  Technology Conference. 1995.
• 6. Systems, S. Overview of COCOMO. 2007 cited
  Available from http//www.softstarsystems.com/ove
  rview.htm.

15
References Cont.

• 7. C. Abts, B.C., S. Devnani-Chulani, E.
  Horowitz, R. Madachy, D. Reifer, R. Selby, B.
  Steece, COCOMO II Model Definition Manual.
  Technical report, Center for Software
  Engineering, USC. 1998.
• 8. Albrecht, A., Function Points A New Way of
  Looking at Tools. 1979.
• 9. Parametric Cost Estimating Handbook. US Dept.
  of Defense, Washington D.C., 1995.
• 10. Agency, D.C.M. DCMA Guidebook - Software
  Acquisition Management. 2007 cited.
• 11. Boehm, B.A., C. Chulani, S., Software
  Development Cost Estimation Approaches - A
  Survey. Annals of Software Engineering, 2000.
  10(1-4) p. 177-205.
• 12. Chris, F.K., An empirical validation of
  software cost estimation models. Commun. ACM,
  1987. 30(5) p. 416-429.
• 13. Sultanodlu, S. Software Measurement, Cost
  Estimation, SLIM, COCOMO. 1998 cited
  Available from http//yunus.hacettepe.edu.tr/sen
  cer/cocomo.html