Aided decision-making through visual analytics systems for large multivariate, spatiotemporal, hierarchical and network data

Sungahn Ko, Purdue University

Abstract

As technologies have advanced, various types of data been produced in science and industry, and extracting actionable information for making effective decisions becomes increasingly difficult for analysts and decision makers. The main reasons causing such difficulty are two-fold; 1) the overwhelming amount of data prevents users understand the data in the exploration, and 2) the complexity of the multiple data characteristics (multi-variate, spatial, temporal or/and networked) needs an integrated data presentation for finding any pattern, trend, anomaly for decision-making. To overcome the analysts' information overload and enable effective visual presentation for efficient analysis and decision making, an interactive visual exploration and analysis environment is needed since traditional machine learning and big data analytics alone are insufficient. This dissertation presents an integrated visual analytics and data exploration framework that allows users to effectively explore and analyze multi-variate, spatio-temporal, and network data. We design and incorporate new visual representations and visualization techniques and apply our work to real world data sets, including sales data and economic impact data, as well as flight delay data across US airports. Our framework helps users to answer hypotheses by visualizing a large table data with pixel-based visualization. In order to present a maximum amount of multi-variate data to a given available screen space, our framework extends the pixel-based matrices and provides interaction methods, including a magnification lens. In addition, our framework incorporates forecasting algorithms (e.g., ARIMA) to present trends on the data of interest. In this way, our framework effectively supports users who constantly explore and analyze the business sector data (e.g., market share analysts) and need to verify future trends and anomalies. In order to enable effective exploration of high-dimensional multi-variate network data exploration, we design and implement two novel visual representations, Petals and Threads. Then, this dissertation describes case studies on US airport flight delay network to demonstrate how our framework can be applied to real world problems. Lastly, for evaluation we present a user study result of the petals and threads and feedback from a domain expert.

Degree

Ph.D.

Advisors

Ebert, Purdue University.

Subject Area

Computer Engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS