Taxonomies can serve as a valuable tool to capture dimensions and characteristics of data analytics solutions in a structured manner and thus create transparency about different design options of the technical solution space. However, previous taxonomic approaches often remain at a purely descriptive level without leveraging morphological structures to investigate the mechanisms between different combinatorial options given in data analytics pipelines. To this end, we propose a taxonomic evaluation approach to evaluate and construct the technical core of analytical information systems more systematically. Specifically, we present a rough guidance model consisting of four steps, which we subsequently instantiate with two application scenarios from the fields of industrial maintenance and predictive business process monitoring. In this way, we demonstrate how taxonomic frameworks can guide the creation of structured evaluation studies to consider the construction and assessment of data analytics pipelines in a multi-perspective and holistic manner. Our approach is sufficiently generic to be applied to various domains, scenarios, and decision support tasks.

分类法可以作为一种有价值的工具，以结构化的方式捕获数据分析解决方案的维度和特征，从而为技术解决方案空间的不同设计选项创建透明度。然而，以前的分类方法通常停留在纯粹的描述性水平，没有利用形态结构来研究数据分析管道中给出的不同组合选项之间的机制。为此，我们提出了一种分类评估方法，以更系统地评估和构建分析信息系统的技术核心。具体来说，我们提出了一个由四个步骤组成的粗略指导模型，我们随后用工业维护和预测业务流程监控领域的两个应用场景对其进行了实例化。这样，我们展示了分类框架如何指导结构化评估研究的创建，以多视角和整体的方式考虑数据分析管道的构建和评估。我们的方法足够通用，可以应用于各种领域、场景和决策支持任务。