Effects of Parameter Uncertainties on Software Development Effort Estimates
Frederic D. Powell · 1990
Abstract : The Constructive Cost MOdel (COCOMO) for software cost estimation uses inputs of size in lines of code and 15 Development Effort Multipliers (DEMs) that calibrate the estimate to the environment. This study examines the effects of uncertainties, expressed as probability densities of size and DEMs, on COCOMO effort estimates for software developments consisting of more than one Computer Software Configuration Item (CSCI). A Taylor series is used to find the mean and standard deviation of effort in terms of the means and standard deviations of the sizes and the DEMs of the CSCIs. The Central Limit Theorem then enables the evaluation of probability density and cumulative distribution functions for effort. Conditions under which this method is valid are demonstrated.