Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
461990 | Journal of Systems and Software | 2012 | 9 Pages |
The size of software project teams has been considered to be a driver of project productivity. Although there is a large literature on this, new publicly available software repositories allow us to empirically perform further research. In this paper we analyse the relationships between productivity, team size and other project variables using the International Software Benchmarking Standards Group (ISBSG) repository. To do so, we apply statistical approaches to a preprocessed subset of the ISBSG repository to facilitate the study. The results show some expected correlations between productivity, effort and time as well as corroborating some other beliefs concerning team size and productivity. In addition, this study concludes that in order to apply statistical or data mining techniques to these type of repositories extensive preprocessing of the data needs to be performed due to ambiguities, wrongly recorded values, missing values, unbalanced datasets, etc. Such preprocessing is a difficult and error prone activity that would need further guidance and information that is not always provided in the repository.
► We review the existing literature connecting software team composition and productivity. ► We analyse existing knowledge by contrasting with the data in the international ISBSG database. ► We discuss the statistical correlations between team size, effort, productivity and project and how they do not always confirm what one would expect from the literature or intuition. ► We present threats to validity and the issues and pitfalls in data cleaning required for this kind of studies