Graduate Thesis Or Dissertation
 

Efficient Algorithms for Robust Spatiotemporal Data Analysis

Public Deposited

Downloadable Content

Download PDF
https://ir.library.oregonstate.edu/concern/graduate_thesis_or_dissertations/w0892j81t

Descriptions

Attribute NameValues
Creator
Abstract
  • Many large-scale data analysis applications involve data that can vary over both time and space. Often the primary goal of analyzing spatiotemporal data is identifying trends, movements, and sudden changes with respect to time, location, or both. This can include a variety of applications in economics (housing prices, unemployment, job movement, etc), city planning (traffic, power consumption, resource allocation, etc), and ecology (migration patterns, species variety, habitat change, etc). Like many domains, one of the major challenges of spatiotemporal data is dealing with noise and missing or untrustworthy observations. These uncertainties make it difficult to ascertain the distinct roles that changes in time and location have on the data. To this end, I have developed two different approaches for dealing with data uncertainty in different spatiotemporal applications. The first approach, dubbed the Quantile Scan algorithm, makes use of quantile regression to more accurately identify anomalous regions in the data. The flexibility of this framework allows ‘anomalies’ to be defined with respect to any quantile of interest. I develop a version of the Quantile Scan algorithm for analyzing spatial, and spatiotemporal data. The second approach is a unique variation of Collective Graphical Models (CGMs) to incorporate multiple views of the data. This multiview model learns and leverages shared information between the views to better compensate for missing observations. Both the Quantile Scan and Multiview CGM algorithms improve accuracy and robustness on noisy data without sacrificing runtime. The speed and accuracy of these models is demonstrated on a variety of synthetic and real-world datasets, compared against existing algorithms.
License
Resource Type
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Committee Member
Academic Affiliation
Rights Statement
Funding Statement (additional comments about funding)
  • Funded in part by NSF grant CCF-1521687.
Publisher
Peer Reviewed
Language
Embargo reason
  • Pending Publication
Embargo date range
  • 2021-09-15 to 2022-10-16

Relationships

Parents:

This work has no parents.

In Collection:

Items

Thumbnail Title Date Uploaded Visibility Actions
file details MooreTravisW2021.pdf 2021-09-15 Public Download