View Item 
        •   Utrecht University Student Theses Repository Home
        • UU Theses Repository
        • Theses
        • View Item
        •   Utrecht University Student Theses Repository Home
        • UU Theses Repository
        • Theses
        • View Item
        JavaScript is disabled for your browser. Some features of this site may not work without it.

        Browse

        All of UU Student Theses RepositoryBy Issue DateAuthorsTitlesSubjectsThis CollectionBy Issue DateAuthorsTitlesSubjects

        A Z-score for Open Source Projects

        Thumbnail
        View/Open
        thesis-zscore.pdf (16.44Mb)
        Publication date
        2013
        Author
        Syed, S.A.S.
        Metadata
        Show full item record
        Summary
        This research is focused towards the prediction of Free Libre Open Source (FLOSS) project failure and non failure. By using the characteristics of open source projects, referred to as determinants, we strive to create a classification model that predicts project failure and additionally, project non-failure. Examples of project characteristics are the number of developers, the number of downloads, the number of releases and other indicators that relate to success or failure. In order to arrive at such classification model, we adopted the method to predict corporate bankruptcy by E. Altman, also known as the Z-score in economic context. The research method employed by Altman provides us with the necessary steps towards a classification method for FLOSS project failure. That is, creating a sample based on a priori groupings (failed and non-failed projects), possibly one year prior to the event, and performing multiple discriminant analysis to create a linear function that best discriminates between the chosen groups. This enables project administrators, or other stakeholders of open source projects, to assess their project outcome and possibly steer it into a more successful outcome. The Z-score model for open source projects, as we have named it, is able to predict 65% of the cases correctly. Meaning, any new project can be classified with 65% accuracy for its outcome, being a failure or non-failure. Furthermore, the model works best for predicting open source project failure, as approximately 70% of the cases can be correctly predicted. Although we believe that essential indicators for open source success or health are hard to measure in numeric values, making the classification model only a reflection of data that can be operationalized, we believe this is first step towards a concrete model to assess open source projects.
        URI
        https://studenttheses.uu.nl/handle/20.500.12932/13789
        Collections
        • Theses
        Utrecht university logo