Weber, ThomasHußmann, HeinrichMühlhäuser, MaxReuter, ChristianPfleging, BastianKosch, ThomasMatviienko, AndriiGerling, Kathrin|Mayer, SvenHeuten, WilkoDöring, TanjaMüller, FlorianSchmitz, Martin2022-08-312022-08-312022https://dl.gi.de/handle/20.500.12116/39260Machine Learning systems are, by now, an essential part of the software landscape. From the development perspective this means a paradigmatic shift, which should be reflected in the way we write software. For now, the majority of developers relies on traditional tools for data-driven development, though. To determine how research into tools is catching up, we conducted a systematic literature review, searching for tools dedicated to data-driven development. Of the 1511 search results, we analyzed 76 relevant publications in detail. The diverse sample indicated a strong interest in this topic from different domains, with different approaches and methods. While there are a number of common trends, e.g. the use of visualization, in these tools, only a limited, although increasing, number of these tools has so far been evaluated comprehensively. We therefore summarize trends, strengths and weaknesses in the status quo for data-driven development tools and conclude with a number of potential future directions this field.enLiterature ReviewSoftware DevelopmentToolsMachine LearningData-Driven DevelopmentText/Conference Paper10.1145/3543758.3543779