Software parametric models

A parametric model is a set of related mathematical equations that incorporates variable parameters. A scenario is defined by selecting a value for each parameter. Software project managers use software parametric models and parametric estimation tools to estimate their projects' duration, staffing and cost.

In the early 1980s refinements to earlier models, such as PRICE S and SLIM, and new models, such as SPQR, Checkpoint, ESTIMACS, SEER-SEM or COCOMO and its commercial implementations PCOC, Costimator, GECOMO, COSTAR and Before You Leap emerged.

One of the earliest was Software LIfecycle Management (SLIM)^[1] SLIM is based on the Putnam model and backed by a database of over 10,000 completed projects from all types of industries all over the world. The database grows by 200–400 validated projects a year.

Costimator by MTI Systems was developed by Thomas Charkiewicz, a former machinist and manufacturing manager who had studied computer-airded manufacturing at the University of Massachusetts. His software models manufacturing costs.

TruePlanning by PRICE Systems introduces a unifying framework to enable the integration of multiple cost models. These cost estimation models support projects involved with application (software) development, Information Technology (IT) asset management, and hardware development. TruePlanning supports third–party cost models such as COCOMO II.

SEER^[2] was designed by Dan Galorath and features a sophisticated user interface to track/reduce risk by making timely, accurate and insightful decisions about the interdependent and often immeasurable costs, schedules and staffing variables.

COSTAR is a commercial implementation by SOFTSTAR systems of COCOMO, encapsulating trade–offs and what-if analysis, plus user–definable cost drivers.

ProjectCodeMeter^[3] (commercial) cost estimation software implements the newer (2009) model Weighted Micro Function Points and the APPW statistical model, as well as the traditional COCOMO 81, COCOMO II 2000, and REVIC models. It offers a semi-automated estimation process.

The prime advantage of these models is that they are objective, repeatable, calibrated and easy to use, although calibration to previous experience may be a disadvantage when applied to a significantly different project.

These models were highly effective for waterfall model, version 1 software projects of the 1980s and highlighted the early achievements of parametrics. As systems became more complex and new languages emerged, different software parametric models emerged that employed new cost estimating relationships, risk analyzers, software sizing, nonlinear software reuse, and personnel continuity.

References