EEE492.19 COCOMO II PostArchitecture COnstructive COst MOdel

1
EEE492.19COCOMO IIPost-Architecture
COnstructive COst MOdel
(Estimation requires detail that is only
available at the post architecture stage)
Royal Military College of Canada Electrical and
Computer Engineering

• Dr. Terry Shepard
• shepard_at_rmc.ca
• 1-613-541-6000 ext. 6031

Major JW Paul Jeff.Paul_at_rmc.ca 1-613-541-6000
ext. 6656
2
What is COCOMO

• COnstructive COst Model
• effort ?(x1, x2,... xn)
• Why algorithmic model?
• Gives a firm basis for planning
• explaining to management, customers that a
  proposal was unrealistic
• Allows for sensitivity analysis
• Gives a plan with which to judge progress

3
COCOMO History

• 1981
• 1995-6
• 1997
• 2000

• The original COCOMO is introduced in Dr. Barry
  Boehm's textbook Software Engineering Economics. 
  This model is now generally called "COCOMO 81
• Early papers describing COCOMO II published
• The first calibration of COCOMO II is released by
  Dr. Boehm, and named "COCOMO II.1997
• The book Software Cost Estimation with COCOMO II
  (Dr. Barry Boehm, et al) is published to document
  how to apply the latest estimation model.  Most
  of the original Software Engineering Economics is
  still applicable to modern software projects.

4
Emerging Extensions - 2000

• Applications Composition The Application Point
  Model
• COPSEMO Phase Schedule and Effort Estimation
• CORADMO Rapid Application Development Estimation
• COCOTS COTS Integration Estimation
• COQUALMO Quality Estimation
• COPROMO Productivity Estimation
• Expert COCOMO Risk Assessment

5
Fundamental Logic
Early Design Post Architecture
The work required to build the system measured in
person-months
The time required to build the system measured in
months
Some measure of the size of the system

• Effort a Size b

• Duration c Size d

A multiplicative scaling factor
constant
An exponential scaling factor

• Staffing Effort/Duration

6
COCOMO II assumptions

• primary cost driver DSI delivered source
  instructions
• aka kSLOC 1000 source LOC
• cost estimates limited to
• defended actives (eg not user training)
• direct charged labour analysts, PM, designers,,,
• not indirect labour secretaries, janitors, higher
  management
• customer developer play well together
• granularity of estimate GIGO
• Factors are based on

Product Factors Platform Factors Personnel
Factors Project Factors
7
Effort

• Measure effort in Person Months
• PMNS a x ( ? EMi ) x Size b
• where
• a 2.94 (calibrated from 161 projects)
• b 1.01 0.01 x S SFj
• EMi - effort multipliers (i 1 to 17 range from
  .75 1.67)
• SFj - exponential scale factors (j 1 to 5
  range from 0 - 5)
• NS - implies nominal schedule

8
Exponential Scale Factors

• PREC Precedentedness
• FLEX Development flexibility
• RESL Architecture/risk resolution
• TEAM Team cohesion
• PMAT Process Maturity

(See handouts)
9
Scaling Factors each range from 0 to 5

• cause an exponential cost increase

10
Scaling Factors
11
Effort Multipliers

• Product Factors
• RELY Reliability Required
• DATA Database Size
• CPLX Product Complexity
• RUSE Required Usability
• DOCU Documentation Needs
• Platform Factors
• TIME Execution time constraints
• STOR Main storage constraints
• PVOL Platform volatility

• Personnel Factors
• ACAP Analyst capability
• PCAP Programmer capability
• APEX Application experience
• PLEX Platform experience
• LTEX Language tool exp
• PCON Personnel continuity
• Project Factors
• TOOL Use of software tools
• SITE Multi-site development
• SCED Req Development schedule

12
Effort Multipliers

• Product factors

13
Effort Multipliers

• Platform factors

14
Effort Multipliers

• Personnel factors

15
Effort Multipliers

• Project factors

16
Effort Multipliers (Post-Architecture)
every 1 mo.
17
Effort Multipliers (Post-Architecture)
18
Calibration

• The values of all of the factors are derived from
  real projects
• context of those projects matters
• The resulting model produces estimates within
  30 of the actuals 52 of the time for effort.
• Calibrating the COCOMO II Post-Architecture
  Model, Bradford Clark, Sunita Devnani-Chulani and
  Barry W. Boehm, International Conference on
  Software Engineering, 1998, pp. 477-480

19
What is the maximum variation?

• PMNS a x ? EMi x Size (1.01 0.01 x S SF
  )

j
0.14 to 346
1.01 to 1.26
for 1000kSLOC
10001.01 1071 10001.26 6900
.14 x 1000 140
346 x 1000 346,000
20
Aside
Exponential Scale Factor
kLOC1.26
Estimated Effort Required
kLOC1.13
kLOC1.01
1000
5000
Where are most University projects on this scale?
Size of Program in kLOC
21
Development Time

• assume nominal schedule
  schedules can be compressed stretched (75 -
  160)
• time TDEV c x (PMNS)d
• where
• c 3.67
• d 0.28 0.2 x b - 1.01
• b 1.01 0.01 x S SFj (j 1 to 5)

22
Wait a sec...
If total effort in person months is
What is this?
TDEV c x (PMNS)d
Turns out there is also an optimal development
time (which is also based on project size and
other factors)
Now can calculate staff required

• Staffing Effort/Duration

Note Staffing curve is not flat
COCOMO II assumes 152 person hours/month
23
Compress/stretch schedules

• What if it has to be faster?

Going faster increases effort
Going slower does nothing?
24
What else can we do?

• Option analysis
• What is the effect of hiring skilled workers?
• What is the effect of contracting out a portion?
• See spreadsheet

25
Example
26
MiniProject

• 20kSLOC
• all factors are nominal except staffing which is
  at multiple locations
• Question if cost per staff is 6000 per month,
  and new facilities will cost 10000 per month, is
  it worthwhile to move to the new facilities?

27
MiniProject

• PMNS 2.94 x 1.25 x 20 (1.16) 118.7 PM
• TDEV 3.67 x (PMNS).31 17 months
• Staffing 8.34 9 pers
• Total cost 17 9 6000 918,000

28
MiniProject

• PMNS 2.94 x 20 (1.16) 94.96 PM
• TDEV 3.67 x (PMNS).31 15 months
• Staffing 6.37 7 pers
• Total cost 15 7 6000 630,000
  150,000

29
Questions?