DASC-PM v1.1 A Process Model for Data Science Projects

Abstract

In February 2020, the first version of a comprehensive process model for data science projects appeared: the Data Science Process Model (DASC-PM). The positive feedback we have received indicates we were able to contribute to the discussion of data science activities that we were hoping for. Over the last two years, the DASC-PM has found its way into practice, book contributions (such as Alekozai et al., 2021), and scientific conferences (such as Schulz et al., 2020). We would like to sincerely thank all the readers who have shared their experiences with us and drawn our attention to the model’s strengths and potential improvements. Of course, special thanks go to those who actively participated in developing the model further. Without them, the path to this Version 1.1 would have been impossible. This version addresses feedback from theory and practice, as well as a few topics we feel strongly about. For example, we have made the document more legible by giving it a more compelling structure and shorter introductory texts. The model itself now more clearly defines the key areas and phases and their characteristics and shows how their interaction can look in various project configurations, including agile ones. We have examined all the terms used in the document with a critical eye and adjusted and standardized them where necessary. To that end, we have also addressed suggestions for a less formal visualization that is more plausible in practice, and—hopefully, at least—made both the document and the actual model more graphically appealing. Since the DASC-PM was created “by many for many,” we felt it was worthwhile to make the overall presentation of the model more accessible, even if it might be a little less scientifically precise. In terms of content, while developing Version 1.1, we focused on the “project order” phase. Important decisions are made and framework conditions are established at the beginning of data science activities. To that end, we are offering a more comprehensive description of the phase and a practical and applicable questionnaire as a concrete basis for both new and experienced users of data science. Just as in Version 1.0, the results should be seen as the aggregate experiences of all the participants of this working group. This English translation of the original German model makes it possible to use it in international projects, more easily supporting the interdisciplinarity that is intrinsic to data science. All the results presented in the DASC-PM are still mostly based on the feedback of a diverse working group and constitute a state of debate that is meant to serve as a stimulus and support but never claims to have the last word in the very active field of data science. We are pleased that this living vitality will continue to motivate us to discuss and modify the DASC-PM and make it available to a wide audience. If you are interested in participating or want to be kept up-do-date about current developments of the model, contact us at the address given below. Elmshorn, Halle (Saale), Hamburg, Krefeld, Mönchengladbach and Stuttgart in June 2022 | The DASC-PM Core Team | Contact: info@dasc-pm.org |