Information

  • Publication Type: Master Thesis
  • Workgroup(s)/Project(s):
  • Date: June 2014
  • Date (Start): 28. June 2014
  • Date (End): 28. June 2014
  • First Supervisor: Eduard GröllerORCID iD

Abstract

In the so called information age, data is widely available. Sources include data collection as a byproduct (e.g., log files on a server, or as a more concrete example, movement profiles of smartphone users) and data generation for particular purposes (e.g., simulation runs, data gathered from sensors). To benefit from this huge amount of available data, the data must be analyzed and relevant information must be extracted. Visual Analytics has become an important approach to identify and extract relevant information from data, especially with big data sets. However, data can also contain erroneous values for different reasons, e.g., caused by defect sensors. In data warehousing projects, transforming and edit the data into a usable state can account up for 80% of the cost and the development time. This diploma thesis focuses on time-dependent data and presents an extension for the existing visual analytics framework VISPLORE, to support the user in the process of data editing. Using plausibility rules, the user can define data checks and imputation strategies. Three different overviews, a data-based overview, a group-based overview, and a rule-based overview provide insight into the structure of implausible data values and the defined plausibility rules. Implausible values can be imputed using the defined imputation strategies and existing visualization techniques are extended to enable the user getting an overview of the modified values. Real-world data is used to demonstrate two use-cases. Limitations of the provided overviews, e.g., scalability for a large number of plausibility rules, are discussed and ideas for future work are outlined.

Additional Files and Images

Additional images and videos

Additional files

Weblinks

No further information available.

BibTeX

@mastersthesis{Moellinger_Christian_IDE2,
  title =      "Interactive Data Editing of Time-Dependent Data in Visual
               Analysis",
  author =     "Christian M\"{o}llinger",
  year =       "2014",
  abstract =   "In the so called information age, data is widely available.
               Sources include data collection as a byproduct (e.g., log
               files on a server, or as a more concrete example, movement
               profiles of smartphone users) and data generation for
               particular purposes (e.g., simulation runs, data gathered
               from sensors). To benefit from this huge amount of available
               data, the data must be analyzed and relevant information
               must be extracted. Visual Analytics has become an important
               approach to identify and extract relevant information from
               data, especially with big data sets. However, data can also
               contain erroneous values for different reasons, e.g., caused
               by defect sensors. In data warehousing projects,
               transforming and edit the data into a usable state can
               account up for 80% of the cost and the development time.
               This diploma thesis focuses on time-dependent data and
               presents an extension for the existing visual analytics
               framework VISPLORE, to support the user in the process of
               data editing. Using plausibility rules, the user can define
               data checks and imputation strategies. Three different
               overviews, a data-based overview, a group-based overview,
               and a rule-based overview provide insight into the structure
               of implausible data values and the defined plausibility
               rules. Implausible values can be imputed using the defined
               imputation strategies and existing visualization techniques
               are extended to enable the user getting an overview of the
               modified values. Real-world data is used to demonstrate two
               use-cases. Limitations of the provided overviews, e.g., 
               scalability for a large number of plausibility rules, are
               discussed and ideas for future work are outlined. ",
  month =      jun,
  address =    "Favoritenstrasse 9-11/E193-02, A-1040 Vienna, Austria",
  school =     "Institute of Computer Graphics and Algorithms, Vienna
               University of Technology ",
  URL =        "https://www.cg.tuwien.ac.at/research/publications/2014/Moellinger_Christian_IDE2/",
}