Predição do evento de turnover em equipes de software livre e de código aberto

• Ano de defesa: 2019
• Autor(a) principal: Giordano, Douglas Montanha
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; Ciência da Computação; UFSM; Programa de Pós-Graduação em Ciência da Computação; Centro de Tecnologia
• Palavras-chave em Português: Sistemas complexos; Engenharia de software; Predição; Turnover; System complex; Software engineering; Predict; CNPQ::CIENCIAS EXATAS E DA TERRA::CIENCIA DA COMPUTACAO
• Link de acesso: http://repositorio.ufsm.br/handle/1/20676

FOSS projects like RAILS, ELIXIR, LINUX among others have a high employee turnover. This turnover happens when people leave the project or new ones come in. When a con- tributor ceases to be active in a project, a series of negative consequences can affect the software. Among some situations are the lack of maintenance in software modules. The- refore, a developer’s exit can be detrimental to the FOSS project. If a way of knowing who will not make the most contributions to the project is offered, managers or core staff may have time to avoid leaving the contributor or lessen the consequences. This study propose a classifier that predicts turnover when it happens. Based on the FOSS data repositories of the GITHUB, the classifier will offer as a response whether the contributor has last in- teracted or will continue in the project. The results of the best classifier, offered a good performance. The precision was 0.83 and the recall of 0.35 for the class of contributors who will continue in the project. On the other hand, the class of contributors who may be inactive had a precision of 0.81 and an recall of 0.98.