THE ESTIMATION OF SOFTWARE SIZE AND EFFORT: AN APPROACH BASED ON THE EVOLUTION OF SOFTWARE METRICS (MEASUREMENT)
Abstract
Estimating the amount of effort required for software development is probably the most important and difficult aspect of any software project. Most current effort estimation models include project size as one of the most important parameters. Using current estimation techniques, early size estimation is of comparable difficulty to early effort estimation. In this thesis an early size estimation approach based on the early estimation of data-structure metrics is presented. This approach is derived from the assumption that programs can be constructed in a structured way such that most data structures can be developed during program design. The feasibility of this approach was demonstrated by a program construction experiment involving small but non-trivial programming tasks. Size estimates using our approach were as good as or significantly better than subjective estimates made by experimental subjects. Another issue addressed in this thesis is the evolution of software metrics during program development. We studied metric evolution to develop a clearer understanding about the development process. Several distinctive patterns of metric evolution were observed and identified in our research. These patterns of metric evolution were found to hinge significantly upon the development strategy employed in the construction process. In addition, the evolution behavior of certain metrics were modeled in our study using simple functions of development time. An effort prediction method based on this modeling of metric evolution is also presented in this thesis. The feasibility of this method was demonstrated by our program construction experiment in which the effort estimates generated using this method were significantly better than the subjective estimates made by the experimental subjects. This research has shown that software metrics can be measured at the end of the design stage and that they can help significantly in size and effort estimation.
Degree
Ph.D.
Subject Area
Computer science
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.