287x Filetype PPT File size 0.88 MB Source: www.cs.vu.nl
Cost estimation, topics
Quantitative models (E = 2.5 KLOC1.05)
Qualitative models (e.g. expert estimation)
Agile cost estimation
Relate cost to development time
SE, Cost estimation, Hans van Vliet, ©2008 2
On productivity
Even if quantitative models are not that good, the
cost drivers of these models learn us about
productivity:
Writing less code helps
Reuse helps
Quality of people is important
Tools help
…
SE, Cost estimation, Hans van Vliet, ©2008 3
Algorithmic models
Base formula: E = a + bKLOCc
C usually is around 1
C > 1: diseconomy of scale
C < 1: economy of scale
This nominal cost is multiplied by a number of cost
drivers (volatility of requirements, amount of
documentation required, CMM level, quality of
people, …)
SE, Cost estimation, Hans van Vliet, ©2008 4
Walston-Felix
One of the early algorithmic models (1977)
Many factors (29 out of 51 projects)
Only three alternatives (high, medium, low) per
factor
Its form influenced many later models
SE, Cost estimation, Hans van Vliet, ©2008 5
COCOMO (COnstructive COst MOdel)
Very well-documented (Boehm book, 1981)
Basic form: E = bKLOCc, where b (~3) and c (1+ε)
depend on the type of project (mode):
Organic: relatively small and well-known
Embedded: inflexible environment with many constraints
Semidetached: somewhere in between
More complex form: takes into account 15
multiplicative cost drivers
SE, Cost estimation, Hans van Vliet, ©2008 6
no reviews yet
Please Login to review.