A Methodology for the Analysis of Programmer Productivity and Effort Estimation within the Framework of Software Conversion.
Abstract
Software conversion is becoming increasingly significant as the inventory of programs increases and as the life cycle of many applications becomes longer. Although some work has been done in the area of software conversion, it has received little research attention since it has only recently become a more frequent occurrence. This research considered two aspects of software conversion and developed a methodology for the statistical analysis of conversion sample data from the ongoing U.S. Air Force Base Level Data Automation Program. The two areas specifically addressed by this research were programmer productivity and effort estimation. Programmer attributes and program characteristics were studied in relation to programmer productivity in software conversion. Models for explaining productivity were constructed and the impact of organization was also considered. Existing applicable models for software conversion effort estimation were examined and their accuracy was evaluated. Environment specific regression models for effort estimation were also constructed. Several statistical and summarizing techniques were considered for the analysis of the conversion sample data. As various aspects of the data were studied, selected statistical techniques emerged as more appropriate. These provided the basis for the methodology formulated and used throughout the data analysis. The Air Force data was utilized in a case study of the application of the methodology. The analysis of conversion programmer productivity revealed that experience, lines of code, a programmer's knowledge of the program, organization, and other factors and attributes impacted productivity.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 1984
- Accession Number
- ADA145757
Entities
People
- J. D. Fernandez
Organizations
- Air Force Institute of Technology