A Methodology for Assessing Schedule Risk in Software Development Estimates
- 86 Downloads
Planning estimates for software development schedules are often obtained using a schedule-estimating equation that expresses the schedule estimate (in calendar months) as a nonlinear function of the total development effort (in staff-months). For example, the COCOMO schedule model  for embedded-mode developments is Schedule = 2.5 (Effort)0.32 We address the problem of measuring the error of uncertainty in the schedule estimate when the error of the effort estimate and the error of the schedule model (given actual effort data) are known.
We illustrate the methodology for two models calibrated to historical cost data.
KeywordsSoftware Development Beta Distribution Effort Estimate Triangular Distribution Nonzero Probability
Unable to display preview. Download preview PDF.
- 2.Funch, P. G., “Recalibration of Basic and Nominal COCOMO Equations to Recent Air Force Acquisitions,” presented at the Third Annual COCOMO Users’ Group Meeting, Pittsburgh, PA, 3–5 November 1987.Google Scholar
- 3.Wolfinger, B. E., “Calibrating COCOMO for Ada,” presented at the Third Annual COCOMO Users’ Group Meeting, Pittsburgh, PA, 3–5 November 1987.Google Scholar
- 4.Pullen, K. W., “Uncertainty Analysis with COCOMO,” presented at the Third Annual COCOMO Users’ Group Meeting, Pittsburgh, PA, 3–5 November 1987.Google Scholar
- 6.“Risk and Uncertainty,” Section IV in Cost Analysis, G. R. McNichols, ed., selected papers from the symposium, “Operations Research in Cost Analysis,” Arlington, VA, 18–20 May 1983; publ. by the Operations Research Society of America, 1984.Google Scholar
- 7.Garvey, P. R., F. D. Powell, “Three Methods for Quantifying Software Development Effort Uncertainty,” pp. 292–306 in Software Risk Management, B. W. Boehm, ed., an IEEE tutorial, publ. by the IEEE Computer Society Press, Washington, DC, 1989.Google Scholar