| This praxis provides a way for a company engaged in software development to study its software development project data, characterize software productivity and product quality for individual projects and collections of software projects, and derive strategic and tactical planning opportunities to improve their 'business of software.' Avionics software project data from thirty-nine (39) completed industry software development projects is collected and analyzed as part of this praxis to quantify the productivity and quality of each software project and the software organization as a whole. A software reliability analysis is performed first to establish the operational quality of the software for each project, the mean time to failure, defect rates, and probable numbers of post-release defects. A statistical analysis follows to study the project datasets using descriptive statistics and tree-based models. Hypothesis test are conducted next; results determine significant findings that answer specific questions regarding software productivity and quality as initially posed by engineering management. Finally, a data envelopment analysis is conducted to determine 'best-in-class' software development projects and to identify the comparative quality of projects that, for whatever reason, were not 'best.' A sensitivity analysis identifies the magnitude of shortfalls within the non-efficient projects. This information is used to demonstrate identifying an improvement course and direction for the software organization based on the data. Using this multi-objective analysis method to simultaneously compare multiple measures of software project performance coupled with more traditional hypothesis testing and other statistical analyses, engineering management can now identify best performing software projects, understand what makes them best, and chart an improvement course for the software organization. Armed with these results, the software manager may identify specific candidate software engineering activities targeting key improvements that the software organization can undertake---activities most likely to improve future software project and organizational software productivity and product quality directly affecting the company's software bottom line. |
