Files in this item
video/mp4IncVisage_ Prog ... -Rapid-Decision-Making.mp4 (3MB)
|(no description provided)||MPEG-4 video|
|Title:||I’ve seen enough: Progressively improving visualizations to support rapid decision making|
On large datasets, generating visualizations can take a long time, delaying the extraction of insights and hampering decision making. Here, we present a sampling-based approach to generate visualizations faster while improving the displayed estimates progressively. Our approach draws samples of data in iterations and can reveal the “salient” features of the visualization quickly while minimizing error, thus enabling rapid and error free decision making. In the attached demo, we show how one can quickly uncover the year-round flight delay patterns, on a dataset of 150 million flights across US, by visualizing a progressively improving heatmap. Starting from a single rectangle, we progressively unveil one interesting feature at each iteration, by dividing one of the rectangles in the previous iteration into four new rectangles. The color density of a rectangle depicts its relative value—the darker the color, the higher the value. For such a large dataset, traditional tools would require several minutes to generate the final, complete visualization. Our approach, on the other hand, reveals the important features within first few iterations (15), requiring only about 10 seconds: higher delay occurs during the holiday season (end of December and early January) and summer months.
(Acknowledgements) Maryam Aliakbarpour, Ha Kyung Kong, Mangesh Bendre, Eric Blais, Karrie Karahalios, Ronitt Rubinfield, Aditya Parameswaran
|Rights Information:||Copyright 2018 Sajjadur Rahman|
|Date Available in IDEALS:||2019-02-18|