Usability of visual data profiling in data cleaning and transformation
Journal article, Peer reviewed
Accepted version
View/ Open
Date
2017Metadata
Show full item recordCollections
- Publikasjoner fra CRIStin - SINTEF AS [5638]
- SINTEF Digital [2381]
Original version
Lecture Notes in Computer Science. 2017, 10574 480-496. 10.1007/978-3-319-69459-7_32Abstract
This paper proposes an approach for using visual data profiling in tabular data cleaning and transformation processes. Visual data profiling is the statistical assessment of datasets to identify and visualize potential quality issues. The proposed approach was implemented in a software prototype and empirically validated in a usability study to determine to what extent visual data profiling is useful and how easy it is to use by data scientists. The study involved 24 users in a comparative usability test and 4 expert reviewers in cognitive walkthroughs. The evaluation results show that users find visual data profiling capabilities to be useful and easy to use in the process of data cleaning and transformation.