1. Based on Chapter 5 of the book [McConnell 2006]
CS 709B Advanced Software Project Management and Development Software Estimation - III
Based on Chapter 5 of the book [McConnell 2006]
Steve McConnell, Software Estimation: Demystifying the Black Art, Microsoft Press, 2006
May 3, 2012

2. Outline Estimate Influences Project Size
Kind of Software Being Developed
Personnel Factors
Programming Language
Other Project Influences

3. Project Size Size can be measured in
Project size is the single most significant contributor to project costs, effort, and schedule
Size can be measured in
Lines of code (LOC)
Function points
Number of requirements, etc.
Organizations don’t pay attention and:
Estimate costs, effort and schedule without not knowing how big the software will be
Do not adjust costs, effort and schedule when the software size is consciously increased

4. Project Size

5. Project Size
In software, larger the groups of people involved, larger the communication overhead

6. Project Size The consequence is software’s diseconomy of scale
Outside software, an economy of scale implies that the bigger you get (e.g., a plant), the smaller the cost per unit produced becomes
In software, on the contrary, the larger the system becomes, the greater the cost per each unit will be

7. Project Size
The diseconomy of scale in smaller projects

8. Project Size
Aggravating factors in larger projects: project conditions that degrade productivity more quickly than average as project size increases

9. Project Size Nominal effort: 10K to 100K LOC: 13x
Worst case growth:
10K to 100K LOC: 17x
10K to 1000K LOC: 300x

10. Project Size
In conclusion, effort scales up exponentially with size (not linearly)
The main implication of diseconomies of scale is that you cannot estimate correctly a new project based on past projects if they differ significantly in size
To compute the impact of diseconmies of scale, use software tools, e.g., the author’s Construx Estimate (free download)
In similar size projects the diseconomies of scale can be however ignored

11. Project Size
About the same size projects are defined as being a factor of 3 from largest to smallest. In such projects you can safely use a ratio-based estimate, such as LOC/staff month

12. Kind of Software

13. Kind of Software Accounting for the industry you are working in:
Use results in Table 5-2 as a starting point. Note ranges are large, with a typical factor of 10 between high and low ends
Use an estimating model such as Cocomo II, but remember not to use to many control knobs
Use historical data from your organization (best approach)
The kind of software you are developing is the second-most significant contributor to project effort and schedule

14. Personnel Factors
Personnel factors can swing a project estimate by as much as factor of 22 (ranked worst on right vs. ranked right on left, on all categories)

15. Personnel Factors
Numerous studies since 1960s have shown 10:1 to 20:1 differences in individual and team performances
Implications:
You cannot accurately estimate a project if you don’t have some idea of who will be doing the work
Most accurate estimations are when you know who specifically will be doing the work that you estimate
Within a particular organization there is not that much variation though (less than 10:1) because both top-tier and bottom-tier developers tend to migrate towards groups of similar skills

16. Programming Language
The programming language used affects the estimates in four ways:
Team’s experience with the language has an impact of about 40% on the overall productivity rate
Some languages generate more functionality per LOC than others (Table 5-3)
The associated tool support and environment can vary effort up to 50% (from best to worst)
Interpreted languages tend to be more productive (about 2 times) than compiled languages

17. Programming Language

18. Other Factors (Cocomo II)
Subjectivity creeping in Cocomo II is due more to “usage failure” than to “method failure”
Cocomo II has conducted the most statistically rigorous analysis of factors incorporated
Cocomo II’s adjustment factors help gain an understanding of different project influences
The method includes Effort Multipliers (EMs)
Table 5-4 contains EMs from “Very Low” (e.g. 1.22) to “Very High” (e.g. 0.81)
The Influence Factor is the ratio Very Low/Very High (e.g., 1.51) and indicates the degree of the factor’s influence on the overall estimate

19. Other Factors (Cocomo II)

20. Other Factors (Cocomo II)

21. Other Factors (Cocomo II)

22. Other Factors (Cocomo II)

23. Other Factors (Cocomo II)