The other main step in gathering size data was selecting the appropriate anchor
point milestones [Boehm 1996] which could act as viable sources of estimates. We choose the Life Cycle Architecture (LCA) and the Re baselined Life Cycle Architecture (RLCA) milestones to gather two sets of estimated size values – one immediately before rebaselining and the other immediately after rebaselining. The reason for choosing the LCA milestone is that it marks the end of the Elaboration phase. Thus estimates available at this stage are much more mature as compared to estimates produced during the Inception phase. The RLCA milestone, which occurs at the beginning of the Construction phase, was chosen since rebaselining of a project may result in the modification of its scope thereby enabling the team to come up with a revised estimate of software size.
Our first set of estimated size values was obtained using the last version of the
Life Cycle Planning (LCP) document [Boehm et al. 2005] produced in the Fall semester. This version was produced at the time of the LCA milestone. The second set of estimated size values were obtained from the LCP of the RLCA package produced in the Spring semester.
Replaced/Superseded by document(s)
This paper presents an analysis of the results of an empirical study
conducted to examine the relationship between the estimated and actual sizes of small real-client projects. Estimates of software size are collected at two different points of the software development life cycle – one immediately before rebaselining the project mid-stream and the other immediately after rebaselining. It is found that estimates made after rebaselining are usually more accurate. The accuracy of estimates based on the category of projects is also
analysed. The results indicate that the correlation between estimated and actual size is much stronger in case of web-based projects vis-à-vis non-web-based projects.