Processes in Data Science Projects

Data science and artificial intelligence have passed the stage of research in the ivory tower over the last years. Applications are not only found in huge enterprises and corporate groups: Many start-up companies were founded, and also small and medium sized enterprises adapt the new technology and take advantage of the capabilities more and more. For many of them, the use of data-based approaches rapidly become a necessity due to the product and service range of the competition or customer expectations. In particular, companies coming from other business sections than information technology face the challenge to implement new and robust data-based solutions. Classical structures and competencies have to be combined with new ones in data science projects, which usually come with high interdisciplinarity. Some aspects of such projects can be done just as in classical projects whereas others have to be slightly adapted and also some completely new arise. Data science process models can assist enterprises by facing these challenges with a structured approach, however most of them focus on the new or technical aspects of such projects or ignore the business context. This paper focuses on the aspect of business processes from data science projects in practice and shows their relevance in several points of time in and around a project’s lifetime. Process-related differences to classical projects are shown and possibilities to take processes into account in an appropriate manner are discussed. Lastly, recommendations are given to cope with processes in the context of data science projects respecting the interplay of processes, humans and technology.