A process model for systematically setting up the data basis for data-driven projects in manufacturing

In the rapidly advancing fields of Artificial Intelligence (AI) and Big Data, creating a robust and high-quality data foundation is a critical requirement for data-driven projects. However, the lack of a standard procedure for ensuring the existence of a sufficient and high-quality data basis often leads to misunderstandings, inefficiencies, and resource waste, resulting in a high risk of project failure. Existing methodologies often presuppose the availability of a data basis, which is a significant challenge, particularly in the manufacturing sector with its diverse and complex data sources. This challenge is further compounded by the interdisciplinary nature of these projects, where domain experts and data scientists with different expertise and vocabularies must collaborate. Addressing this gap, this paper introduces ML-SIPOC, a novel methodology for creating a standardized and high-quality data basis for data-driven projects. ML-SIPOC builds upon the traditional SIPOC analysis from the Six Sigma management system for operational excellence, adapted to meet the unique challenges of data-intensive projects. It provides a structured framework for systematically building a data foundation, facilitating effective communication between domain experts and data scientists. When applied in electronics manufacturing, ML-SIPOC proved efficient in creating a robust data basis, reducing time and cost overheads. This approach minimizes reliance on prior knowledge for data collection, opening up possibilities for broader AI and Big Data applications across various manufacturing sectors. The key innovation of this paper is the introduction of a first-of-its-kind methodology that provides a structured approach to building a data foundation for data-driven decision making in manufacturing.