Brief cosmos Product Description
COSMOS is an estimation tool that runs on PCs with Microsoft Windows 95 or Microsoft Windows NT. Detailed instructions for using COSMOS are given in the COSMOS User’s Guide and the on-line help. Further examples are given in the COSMOS User’s Guide.
COSMOS produces estimates of a software project's effort, schedule, and staffing requirements. COSMOS combines Allan Albrecht’s (IFPUG standard) function point analysis, Barry Boehm's COCOMO model, and Lawrence Putnam’s adaptation of the Rayleigh distribution to produce these estimates. COSMOS combines three landmark software engineering models into one system for sophisticated project forecasting. These models are described below. While the parameters used to tune COSMOS come from principal software engineering references, the knowledgeable user may re-tune COSMOS to align with conditions in his or her development environment.
Cosmos Capabilities and Theory of Operation
COSMOS produces estimates of a software development project’s effort or schedule. In a nutshell, COSMOS works as follows:
A project is sized using standard Function Point Analysis(as described in the International Function Point User’s Group [IFPUG] Counting Practices Manual version 4.0).Function pointsand how they are counted are described in chapter two of this document. An implementation language is selected, and number ofsource lines of code (SLOC)is estimated from the adjusted function point count and the selected programming language using theBackfire Method. While not a part of function point analysis per se, the Backfire Method is used with the function point count for a project, and is built into COSMOS’s implementation of the Function Point Model. The Backfire Method is also described in chapter two.
The SLOC estimate is fed into a selected COCOMOModel, along with cost drivers. The COCOMO Model computeseffort and scheduleestimates. COCOMO estimation is described in chapter three.
The SLOC and schedule are fed into a RayleighModel that produces a“Manpower Buildup Index” (MBI). The MBI reflects the rapidity with which a project is staffed and scheduled. The MBI predicts whether (and with what difficulty) the project can be accomplished within the time frame determined by the COCOMO model. The COSMOS Rayleigh Model is described in chapter four.
The three COSMOS models – Function Point, COCOMO, and Rayleigh – are combined and linked into a project in order to produce estimates of effort, schedule, and staffing based on the project’s size (i.e. functionality) and the development environment’s most influential characteristics.
Cosmos Project Level Estimation
Since function points can be accurately counted early in the software development cycle, COSMOS’s use of function points as the project size input allows for earlier, more accurate estimates than products that require source lines of code as the primary input. COSMOS combines the advantages of function point analysis with the robustness of the statistically calibrated COCOMO and Rayleigh models.
In addition to project-level estimation, COSMOS can produce stand-aloneFunction Point, COCOMO, and Rayleigh models (the three kinds used to construct a project). These stand-alone models can be imported into other projects. They can also be used to support the use of other parametric cost estimation tools. For example, a COSMOS Function Point model can produce a SLOC estimate for use in products that require SLOC as the project size input.
Groups of projects can be collected into a “System”, and the effort and schedule for developing the entire system can be estimated.
