In the age of Industry 4.0 and Cyber-Physical Production Systems (CPPSs) vast amounts of potentially valuable data are being generated. Methods from Machine Learning (ML) and Data Mining (DM) have proven to be promising in extracting complex and hidden patterns from the data collected. The knowledge obtained can in turn be used to improve tasks like diagnostics or maintenance planning. However, such data-driven projects, usually performed with the Cross-Industry Standard Process for Data Mining (CRISP-DM), often fail due to the disproportionate amount of time needed for understanding and preparing the data. The application of domain-specific ontologies has demonstrated its advantageousness in a wide variety of Industry 4.0 application scenarios regarding the aforementioned challenges. However, workflows and artifacts from ontology design for CPPSs have not yet been systematically integrated into the CRISP-DM. Accordingly, this contribution intends to present an integrated approach so that data scientists are able to more quickly and reliably gain insights into the CPPS. The result is exemplarily applied to an anomaly detection use case.