Proceedings: GI 2019

FSelector: Variable Selection Using Visual Features

Tommy Dang (Texas Tech University)

Proceedings of Graphics Interface 2019: Kingston, Ontario, 28 - 31 May 2019

DOI 10.20380/GI2019.07

  • BibTex

    @inproceedings{Dang:2019:10.20380/GI2019.07,
    author = {Dang, Tommy},
    title = {FSelector: Variable Selection Using Visual Features},
    booktitle = {Proceedings of Graphics Interface 2019},
    series = {GI 2019},
    year = {2019},
    issn = {0713-5424},
    isbn = {978-0-9947868-4-5},
    location = {Kingston, Ontario},
    numpages = {9},
    doi = {10.20380/GI2019.07},
    publisher = {Canadian Information Processing Society},
    }

Abstract

Visual representation of large datasets should allow us to focus on essential dimensions when restricted to limited visual space. This paper presents an approach for abstracting multi-dimensional data with a focus on grouping the individual attributes based on visual features (or Scagnostics) such as density, skewness, shape, outliers, and texture. Working directly with these visual characterizations, we propose a prototype, called FSelector, to guide users when interactively exploring high dimensional datasets. In particular, selected (leading) variables are organized in a grid layout, allowing users to rapidly identify interesting pairs of variables and to focus on analyzing the original variables directly.