ICS2000-12-01: Empirically Guided Software Effort Guestimation - P. Johnson, C. Moore, J. Dane, R. Brewer

For several years, we have pursued an initiative called Project LEAP, whose goal is the improvement of individual developers though lightweight,empirical, anti-measurement dysfunction, and portable software engineering tools and methods.During the Fall of 1999, we performed a case study using the LEAP toolkit in a graduate software engineering class. One of the goals of the study was to evaluate the various analytical estimation methods made available by the toolkit. We were curious as to whether a single method would prove most accurate, or whether the most accurate method would depend upon the type of project or the specific developer. To our surprise, we found that,in most cases, the developer-generated ``guesstimates" were more accurate than the analytical estimates.We also found that the PROBE method of the Personal Software Process, perhaps the most widely publicized analytical approach to personal effort estimation, was the sixth most accurate method. Finally, we found that access to a range of analytical estimation methods appeared to be useful to developers in generating their guesstimates and improving them over time.

ICS2000-12-01.pdf — PDF document, 71Kb