A1650
Title: Quantifying the intrinsic data quality of process data
Authors: Chong Dae Kim - TH Koeln (Technische Hochschule Koeln) (Germany) [presenting]
Abstract: In an increasingly digitized world, the role of data in all its forms is essential. This importance aligns with the growing emphasis on data sharing. To allow data consumers to assess the quality of data in advance, it should be recorded in a manner that is easily and clearly reproducible. Following an extensive literature review that did not yield suitable methods, a framework for quantifying the intrinsic data quality of process data is presented in the form of time series. For this purpose, the dimensions of intrinsic data quality presented are individually made mathematically assessable and implemented into a Python framework for testing, validating and producing results. The quantification method presented is validated using measurement data obtained from a gear test rig, accompanied by a systematic evaluation of the dataset. It is demonstrated that the methodology considers various dimensions in addition to accuracy. Nevertheless, score enhancement is achievable through data preprocessing. Further research should explore the application of the methodology across diverse datasets and quantify additional dimensions of data quality.
