Software Cost and Productivity Model

Abstract A software cost and productivity model is summarized. The model is based on 452 software development activities that commenced or completed after January 1, 1996, and are drawn from air, sea, land, and space programs. Software complexity is considered by assigning each data record to one of 32 combinations of operational environment and application domains. In addition, the model captures productivity information based on programming language. The following areas are examined, with comparisons to an earlier (1996) Aerospace Corporation model whenever possible: (1) software development productivity, (2) code growth, (3) apparent minimum and maximum software development schedules as a function of the number of source lines of code, (4) relationships for estimating the number of developer-months needed to produce a given number of lines of code, and (5) approximate data distributions. A key finding is that an apparent increase in software productivity has occurred, with median levels of productivity in two software environment groupings increasing approximately 30%.