A Platform for Machine Learning in Environmental Agencies—Concept with Use Cases, User Roles, System Functions

Given global ecological challenges, environmental agencies need a reliable, data-based basis for decision-making. However, increasing volume and heterogeneity of data makes it difficult to analyze environmental data. Machine learning (ML) offers methodological support for this but has so far only been used to a limited extent in environmental agencies due to a lack of ML expertise. The Simplex4Learning project is developing a ML platform for environmental agencies. This article describes a central development step: based on a case study analysis with three environmental authorities, generalized use cases and a differentiated role concept were derived. The Explorer and Analyst roles as well as specific system functions along the ML process were defined to provide specific support for domain experts and enable them to use ML methods effectively even without in-depth technical knowledge. A system architecture for the development of the ML platform was proposed, which is characterized by the integration of the envVisio method with the Cadenza data analysis software and reusable modules for modeling, inference calculation, and result presentation. Our results contribute to a structured integration of machine learning methods in environmental analyses by environmental agencies.