Combining Automated Analysis with Interactive Visualizations for Spatio-temporal Data Analysis

The Hong Kong University of Science and Technology
Department of Computer Science and Engineering


PhD Thesis Defence


Title: "Combining Automated Analysis with Interactive Visualizations for 
Spatio-temporal Data Analysis"

By

Mr. Dongyu LIU


Abstract

The rapid advances in sensing technologies and large-scale computing 
infrastructures lead to explosive growth in data. Spatio-temporal (ST) 
data, as one of the ubiquitous data types, is increasingly collected and 
extensively studied in various scientific domains such as geology, 
climatology, sociology, and transportation science. This kind of data is 
distinct from other data types due to the simultaneous presence of spatial 
and temporal dimensions, which increases analysis complexity 
substantially. Purely automatic data analysis techniques, therefore, are 
insufficient to handle such complexity immaculately. Humans not only have 
inherently good senses for perceiving space and time, but also have 
creativity, flexibility, and domain expertise. Hence, an appropriate 
manner to involve these humans' traits into the automatic data analysis 
process would be tremendously helpful.

In this thesis, we introduce three novel visual analysis techniques for ST 
data analysis to demonstrate the benefits brought by the combination of 
automatic data analysis techniques with interactive visualizations. 
Firstly, we study how to solve a multi-criteria decision-making problem in 
the spatial-temporal context that involves a vast solution search space. 
We use optimal billboard location selection as our application scenario 
and propose a system, SmartAdP. The system integrates a novel 
visualization-driven data mining model with tailored data index mechanisms 
to facilitate efficient solution formulation and several well-designed 
visualizations to support optimal solution identification. Secondly, we 
investigate how to detect and examine anomalous events in a large number 
of spatial time series. We use air quality analysis as our application 
scenario and present AQEyes, a system that integrates a unified end-to-end 
tunable machine learning pipeline for quick anomaly identification and a 
set of novel visualizations for efficient exploration and examination of 
air quality dynamics and anomalous events. Thirdly, we research how to 
quickly identify spatio-temporal patterns hidden within subsets of 
large-scale multidimensional spatio-temporal datasets. We propose a novel 
tensor-based algorithm to allow automatically slicing the data into 
homogeneous partitions and extracting the latent patterns in each 
partition for comparison and visual summarization. Based on the algorithm, 
we further develop a system, TPFlow, to support a top-down, 
human-steerable, and progressive partitioning workflow for level-of-detail 
multidimensional ST data exploration.

The effectiveness and usefulness of the above techniques are validated 
through case studies on real-world datasets and interviews of domain 
experts. Notice that the proposed techniques are not limited to the 
presented example application scenarios. They can be easily adapted to 
other applications with similar problems as well.


Date:			Wednesday, 21 August 2019

Time:			10:00am - 12:00noon

Venue:			Room 3494
 			Lifts 25/26

Chairman:		Prof. Alexis Lau (ENVR)

Committee Members:	Prof. Huamin Qu (Supervisor)
 			Prof. Pedro Sander
 			Prof. Yangqiu Song
 			Prof. Zhongming Lu (ENVR)
 			Prof. Cláudio Silva (New York University)


**** ALL are Welcome ****