Measuring Manufacturing Test Data Analysis Quality

Manufacturing test data volumes are constantly increasing. While there has been extensive focus in the literature on big data processing, less focus has existed on data quality, and considerably less focus has been placed specifically on manufacturing test data quality. This paper presents a fully automated test data quality measurement developed by the authors to facilitate analysis of manufacturing test operations, resulting in a single number used to compare manufacturing test data quality across programs and factories, and focusing effort cost-effectively. The automation enables program and factory users to see, understand, and improve their test data quality directly. Immediate improvements in test data quality speed manufacturing test operation analysis, reducing elapsed time and overall spend in test operations. Data quality has significant financial impacts to businesses [1]. While manufacturing cost models are well understood, data quality cost models are less well understood (see Eppler & Helfert [2] who review manufacturing cost models and create a taxonomy for data quality costs). Kim & Choi [3] discuss measuring data quality costs, and a rudimentary data quality cost calculation is described in [4]. Haug et al. [5] describe a classification of costs for poor data quality, and while they do not provide a cost calculation, they do define optimality for data quality. Laranjeiro et al. [6] have a recent survey of poor data quality classification. Ge & Helfert [7] extend the work in [2], and provide an updated review of data quality costs. Test data is specifically addressed in the context of data processing in [8]. Big data quality efforts are reviewed in [9, 10]. Data quality metrics are discussed in [11], and requirements for data quality metrics are identified in [12]. Data inconsistencies are detailed in [13], while categorical data inconsistencies are explained in [14]. In the current work, manufacturing test data quality is directly correlated to the speed of manufacturing test operations analysis. A measurement for manufacturing test data quality indicates the speed at which analysis can be performed, and increases in the test data quality score have precipitated increases in the speed of analysis, described herein.