CS510 Midterm I
Fall 2011
3 questions, 100 points
October 3, 2011
Name: ____________________
Student ID: _____________________
Email: ____________________
DEN Student: Yes/No
Question 1 (40)
Question 2 (30)
Question 3 (30)
Total (100)
Question 1. (40 points)
You have a staff consisting of half Very High Capability Analysts and Programmers and half
Very Low. Your next project consists of half Very High Complexity software and half Very Low.
The nominal person-months for the software independent of complexity and capability is 100
person-months each for the Very High and Very Low Complexity software. Compute the
estimated person-months for the project using the following three strategies and indicate the best
choice:
a. Assign the Very High Capability personnel to the Very High Complexity software,
Assign the Very Low Capability personnel to the Very Low Complexity software;
b. Assign the Very High Capability personnel to the Very Low Complexity software,
Assign the Very Low Capability personnel to the Very High Complexity software,
c. Assign all of the software to an equal mix of Very High Capability and Very Low
Capability personnel (average rating of Nominal);
Solution:
Case a. VH CPLX: (100PM)*(1.34)*(0.71)*(0.76) =72.3PM (6 points)
VL CPLX: (100PM)*(0.73)*(1.42)*(1.34) =138.9PM (6 points)
Total Effort=72.3+138.9=211.2PM (2 points)
Case b. VH CPLX: (100PM)*(1.34)*(1.42)*(1.34) =255.0 PM (6 points)
VL CPLX: (100PM)*(0.73)*(0.71)*(0.76) =39.4 PM (6 points)
Total Effort=255.0+39.4=294.4PM (2 points)
Case c. VH CPLX: (100PM)*(1.34)*(1.0) =134PM (3 points)
VL CPLX: (100PM)*(0.73)*(1.0) = 73PM (3 points)
Total Effort=134+73=207PM (2 points)
Case c is the best (4 points)
Question 2. (30 points)
Consider the Reliability Level Life Cycle Cost trade off analysis results in the bottom row of
Table 2.43 in the COCOMOII book. The Total Ownership Cost (TOC) is equal to the Life Cycle
Cost (LLC) plus the Cost of Defects for each category of software: High Value, Medium Value,
and Low Value software. Assuming the software size is 100KSLOC, and LLC per SLOC of the
software is the same for each value level (i.e., LLC divided by 100KSLOC); compute the TOC
per SLOC for each Reliability level if the cost of defects per SLOC is as follows for the three
value levels.
Reliability Level Cost of Defects Per SLOC
Value Level
VL L N H VH
High $100 $80 $60 $40 $20
Medium $50 $40 $30 $20 $10
Low $20 $16 $12 $8 $4
Compute the TOC/SLOC as a function of Reliability level for each value level, and indicate the
best choice for each value level.
Solution:
Reliability VL L N H VH
Level
LCC $15488K $14168K $13200K $13464K $15244K
LCC/SLOC $154.88 $141.68 $132.00 $134.64 $152.24
(6 points)
Defect $100 $80 $60 $40 $20
High
Cost/SLOC
Value
TOC/SLOC $254.88 $221.68 $192.00 $174.64 $172.24
(6 points)
Defect $50 $40 $30 $20 $10
Medium Cost/SLOC
Value TOC/SLOC $204.88 $181.68 $162.00 $154.64 $162.24
(6 points)
Defect $20 $16 $12 $8 $4
Cost/SLOC
Low Value
TOC/SLOC $174.88 $157.68 $144.00 $142.64 $156.24
(6 points)
For High Value, the best is VH RELY rating; (2 points)
For Medium Value, the best is H RELY rating; (2 points)
For Low Value, the best is H RELY rating; (2 points)
Question 3 (30 points)
Indicate true (T) or false (F) and supporting rationale for each statement (1 point each for
the answer, and 4 points for the rationale)
1. _____ If you do not actively attack the risks, the risks will actively attack you.
Rationale:
True, if you do not mitigate the risk, the risk will turn into a problem.
2. _____ DATA can be determined by using the size of data storage (KB)
Rationale:
False, DATA is determined by the ratio of bytes in the testing database to SLOC in the
program
3. _____ If you are considering developing a COTS-Based system, it is better to use a bottom-
up development, rather than a top-down development.
Rationale:
False, to develop a CBS, you should not start with fixed requirements, rather you should start
with win conditions or specification of COTS.
4. _____ Cowboy programming generally yields very low defect density.
Rationale:
False, cowboy programming or staying 72 hours to code before delivery time generally
creates a lot of defects.
5. _____ Productivity is the best measure to evaluate a performance of a team member.
Rationale:
False, experiments on Page 33 of lecture People and values shows that productivity alone is
not a good way to measure the performance.
6. _____ It is true that all win-lose negotiation result will eventually lead to lose-lose situation
for the project.
Rationale:
True, by principles of negotiation, if one stakeholder is losing, it will lead to lose-lose
situation. For example, if a maintainer prefer C#, but the developer develops by using java, it
yields win-lose situation, but finally no one will maintain it, so it is a lose-lose situation.