R&D Projects in manufacturing are risky! Money must be invested even though the outcome is hard to be determined in advance and possible earnings reveal themselves many months or even years after the project has been finished. Project control often looks for KPIs measuring the state of the project at a given point in time. This approach neither looks ahead nor does it allow root cause analysis. My presentation describes a pragmatic approach of deriving project specific indicators out of the project data base and detecting correlations of these to the field issues of the product the project generates. This allows the detection of field issues in the very early state at which the product is just about to be developed. Furthermore, the power of text analytics is exploited to analyze the R&D project’s descriptions by finding custom entities and text topics. This allows the desired root cause analysis. The combined analysis of structured and unstructured data illustrates the benefit of data science to a small sampled business relevant use case.
Dr. Max Köhler, Data Scientist, Continental AG