COCOMO II Drivers
COCOMO II drivers are the ratings that adjust the model’s estimate: five scale factors, which set the exponent on size and so matter more the larger the system, and seven or seventeen effort multipliers, which scale the result up or down. This page gives their values; the equations are on COCOMO.
1. Scale factors
Each factor is rated from Very Low to Extra High, the five values are summed, and the exponent is E = 0.91 + 0.01 × ΣSF [1]:
| Scale factor | Very Low | Low | Nominal | High | Very High | Extra High |
|---|---|---|---|---|---|---|
| PREC precedentedness | 6.20 | 4.96 | 3.72 | 2.48 | 1.24 | 0.00 |
| FLEX development flexibility | 5.07 | 4.05 | 3.04 | 2.03 | 1.01 | 0.00 |
| RESL architecture / risk resolution | 7.07 | 5.65 | 4.24 | 2.83 | 1.41 | 0.00 |
| TEAM team cohesion | 5.48 | 4.38 | 3.29 | 2.19 | 1.10 | 0.00 |
| PMAT process maturity | 7.80 | 6.24 | 4.68 | 3.12 | 1.56 | 0.00 |
PMAT follows the SW-CMM: Very Low and Low split Level 1, Nominal is Level 2, and so on up to Level 5.
The table divides into what a manager inherits and what a manager chooses. PREC and FLEX are “largely intrinsic to a project and uncontrollable”, while “The next three scale factors identify management controllables by which projects can reduce diseconomies of scale by reducing sources of project turbulence, entropy, and rework” [1]. One level on one factor moves effort by about 6%. Across the whole range the effect is large: at 100 KSLOC, every factor at Extra High gives 194 person-months and every factor at Very Low gives 832 — a factor of about four from the exponent alone.
2. Effort multipliers
Post-Architecture uses 17 multipliers in four groups, which can be rated per module, except SCED [1]:
| Group | Multipliers |
|---|---|
| Product | RELY required reliability · DATA database size · CPLX product complexity · RUSE developed for reusability · DOCU documentation match to life-cycle needs |
| Platform | TIME execution time constraint · STOR main storage constraint · PVOL platform volatility |
| Personnel | ACAP analyst capability · PCAP programmer capability · PCON personnel continuity · APEX applications experience · PLEX platform experience · LTEX language and tool experience |
| Project | TOOL use of software tools · SITE multisite development · SCED required development schedule |
Early Design, used before the architecture exists, merges them into seven: PERS (ACAP, PCAP, PCON), RCPX (RELY, DATA, CPLX, DOCU), RUSE, PDIF (TIME, STOR, PVOL), PREX (APEX, PLEX, LTEX), FCIL (TOOL, SITE) and SCED.
SCED is the one a sponsor changes most often. Schedules of 75%, 85%, 100%, 130% and 160% of nominal multiply effort by 1.43, 1.14, 1.00, 1.00 and 1.00: compression costs effort, and “Stretch-outs do not add or decrease effort” [1].
3. Example: rating a 50 KSLOC booking platform
Example. A team is to build a fictional clinic-booking platform of about 50 KSLOC. It resembles earlier products (PREC Nominal, 3.72), the client allows some freedom (FLEX High, 2.03), architecture risks are partly resolved (RESL Nominal, 4.24), the team works well together (TEAM High, 2.19), and the organisation is at CMM Level 2 (PMAT Nominal, 4.68).
ΣSF = 16.86, so E = 1.0786. With every multiplier at Nominal, effort is 2.94 × 50^1.0786 ≈ 200 person-months, and the schedule ≈ 19.3 months with about 10 people. Asking for the platform in 75% of that time costs 1.43 × 200 ≈ 286 person-months in about 14.5 months.
Read TEAM and PMAT one column to the right — 1.10 and 3.12 — and ΣSF drops to 14.21, E to 1.0521, and effort to about 180 person-months: 11% less from two misread cells — the 6%-per-level rule at work.
4. When to rate which, and the main pitfall
Use the seven Early Design drivers while the architecture is still open, and the seventeen once it exists [1]. The pitfall is treating ratings as facts. Each is a judgment, and a project can tune them until the expected number appears — using the model “as a disguise for expert estimation” [2]. Record why each rating was chosen.
How solid is this?
- Where it comes from. Every value here is from the COCOMO II.2000 manual — the scale-factor table, the SCED multipliers and the Early Design groupings — calibrated on 161 affiliate projects [1].
- How to read the range. 194 versus 832 person-months is the manual’s extreme case, every factor at one end; the 6% per level is the usable figure.
- What we do not hold. The rating values of the sixteen multipliers other than SCED are in the manual’s §3.2 and are not reproduced here. The worked example is computed from the manual’s constants, not taken from a project.
Acknowledgments
This page adapts material from lectures by Eduardo Miranda and David Root [3] on software project management.
References
- Center for Software Engineering, USC,COCOMO II Model Definition Manual. Center for Software Engineering, University of Southern California, 2000.
- M. Jørgensen, B. Boehm, and S. Rifkin,“Software Development Effort Estimation: Formal Models or Expert Judgment?,” IEEE Software, vol. 26, no. 2, pp. 14–19, Mar. 2009, doi: 10.1109/MS.2009.47.
- D. Root,Managing Software Development. Carnegie Mellon University / Innopolis University lecture series, 2014.
Disclaimer: AI is used for text summarization, polishing and explaining. Authors have verified all facts and claims. In case of an error, feel free to file an issue.