[Defense] Density based Frameworks for Spatio-temporal Change Analysis
Wednesday, November 16, 2022
9:00 am - 10:30 am
In
Partial
Fulfillment
of
the
Requirements
for
the
Degree
of
Doctor
of
Philosophy
Karima
Elgarroussi
will
defend
her
dissertation
Density
based
Frameworks
for
Spatio-temporal
Change
Analysis
Abstract
Analyzing change in spatio-temporal data is important for many different domains such as biology, ecology, meteorology, medicine, transportation, and forestry. In order to model these data, two challenges must be addressed: first, the joint analysis of spatial and temporal components, and second, the influence of collocated spatio-temporal objects on one another. Density functions, on the other hand, have proven valuable tools in data mining. However, spatio-temporal data mining techniques that employ density estimation functions are still in the early stages. In this dissertation, we develop and implement density-based frameworks for spatio-temporal change analysis. First, we introduce density-based spatio-temporal data analysis for emotion mapping and emotion change analysis. Our proposed approach segments first the input dataset into batches based on a fixed-size time window. Next, by generalizing existing kernel density estimation techniques, each batch is transformed into a weighted continuous function that takes positive and negative values. After obtaining spatial clusters using a contouring algorithm, an emotion change graph with nodes representing spatial clusters and edges representing temporal relationships between spatial clusters is generated and mined. The framework was successfully applied to tweets collected in the state of New York; the experimental results show that the framework can effectively discover interesting spatio-temporal patterns in tweets and analyzes how emotion change over time. Using density functions as a model for distance decay, we developed, as the second theme of this research, a novel density-based collocation mining framework that uses collocation measures to assess the strength of a given collocation pattern in a particular location, relying on non-parametric density estimation techniques. This approach is generalized to 3D spatio-temporal space to mine spatiotemporal collocation patterns using two methods, the 3D space-time density functions, and the batch-based approach. Our experimental evaluation using the NYC real-world crime data set has demonstrated that our approach provides more insights compared to the traditional collocation methods.
Wednesday,
November
16,
2022
9:00AM
CT
Online
via
MS
聽
Dr. Christoph F. Eick, dissertation advisor
Faculty, students and the general public are invited.
