Online Program

Return to main conference page
Saturday, May 19
Data Visualization
Visualizing Complex Data
Sat, May 19, 8:30 AM - 10:00 AM
Grand Ballroom F
 

Quantitative Evaluation of Manufacturing Visualization via Data Fusion (304565)

Presentation

*Xiaoyu Chen, Virginia Tech 
Ran Jin, Virginia Tech 

Keywords: Data fusion, Electroencephalogram, Eye tracking, Quantitative evaluation, Visualization

Heterogeneous data and complex analytics results make it cognitive challenging for manufacturing users (e.g., engineers and operators) to obtain information and insights in collaboration with manufacturing data analytics. Traditionally, a user may spend majority of his/her time to understand data sets and analytics results, and leave a short period of time for decision-making. It makes the human-machine collaboration inefficient. On the other hand, new visualization techniques and platforms provide graphical representations of information, with an objective to reduce cognitive workloads and enhance insights. But a quantitative visualization evaluation method is still lacking in the literature to provide real-time evaluation for visualization improvement. As a result, visualization contents and designs cannot be optimized in a timely manner. Motivated by manufacturing visualization needs, we proposed a data fusion method by collecting data from Electroencephalogram (EEG), eye tracker and user logs (i.e., mouse movements and events) when users are browsing visualization systems. Statistical, morphological, and time-frequency features of the EEG signals, area of interest-related features of eye movements and user logs are extracted. A Lasso regression is adopted to select significant variables which contribute to high complexities, and accurately predict subjective ratings (i.e., the perceived task complexities) of visualization designs. A case study of 15 participants performing 11 predefined visual searching tasks based on three visualization designs is modeled to demonstrate the accuracy and effectiveness of the proposed evaluation methodology.