Προτάσεις για διαδραστική οπτικοποίηση χωρικών - κατηγορικών δεδομένων

Ρούτση, Θεόδωρος; Routsi, Theodoros

Recommendations for Interactive data visualization of spatial - categorical data

Master Thesis

Author

Ρούτση, Θεόδωρος

Routsi, Theodoros

Date

2024-01

Abstract

Data visualization plays a crucial role in modern businesses and organizations. Nowadays, the data volumes are highly increased and the analysts need to manage and visualize all this data to find out trends or insights that may help their research to advance. In order to help analysts visualization recommendation systems are broadly used. These systems offer a collection of capabilities that aim at the visualization of the data contained in enormous datasets much faster than the analysts should have done by examining every variable by themselves, supporting storytelling and helping them to discover many insights. So, this master thesis delves into the world of visualization recommendation systems, providing an extensive overview of the different types of systems available, implementation best practices, and methods for evaluating their effectiveness. Notably, our investigation addresses a recognized gap in Visualization Recommendation Systems (VRSs), emphasizing the absence of systems adept at guiding users toward areas of interest for exploration and in-depth study. Despite the prevalence of tools for geospatial data analysis, existing platforms often lack the functionality to provide directed recommendations on regions worthy of examination, a deficiency our research seeks to rectify. So, in this thesis, we focused on the recommendation of views representing areas through a spatial partitioning scheme, knowing this is a flexible way of representing spatial and geographical data. We used hexagons as a grid, in order to analyze the behavior of the grid itself as representing the behavior of all points underneath instead of every individual point. This thesis covers the innovation of recommending and visualizing hexagons, providing readers with a comprehensive understanding of the potential applications of this new format. Specifically, our contribution is the creation of a visualization recommendation approach that mostly supports geographical data, using a hexagonal grid for studying the behavior of the points underneath in three different use cases. This approach solves two major problems; (a) there are not many existing VRSs that emphasize on the recommendation of geographical areas and (b) the existing VRSs usually recommend specific neighborhoods, cities regions etc. instead of areas that are not depended on the geopolitical boarders, like for instance an area that a part of it belongs to a city and the other part of it belongs to another city. We call these use cases our approach conducts “methods” and they are respectively named as “Global Reference View and Global Target Views”, “Local Reference View and Local Target Views” and “Local Reference View and Local Target Views with Spatial Distance”. The first one returns candidate interesting hexagons defined through all the geographical areas a dataset contains, while the other two methods are related to the study of smaller geographical areas of the dataset defined by the user. The difference between the second and the third method relates to the interestingness the user shows in the spatial distance among the location he wants to study and the hexagons around it. Overall, this thesis is a valuable resource for anyone seeking to gain a deeper understanding of visualization recommendation systems and their applications in modern businesses and organizations. The inclusion of hexagonal visualization adds an innovative and powerful tool to the visualization toolbox, opening up new possibilities for exploring and analyzing complex data sets. Furthermore, we conducted a “User Study” evaluation for our approach and since the results of the evaluation were shown as positive, we believe this is a succeeded starting point for creating a VRS in the future that will effectively and efficiently help analysts in their studies.

Postgraduate Studies Programme

Κυβερνοασφάλεια και Επιστήμη Δεδομένων

Department

Σχολή Τεχνολογιών Πληροφορικής και Επικοινωνιών. Τμήμα Πληροφορικής

Number of pages

Language

English

URI

https://dione.lib.unipi.gr/xmlui/handle/unipi/16207
http://dx.doi.org/10.26267/unipi_dione/3629

Collections

Τμήμα Πληροφορικής

Show full item record

Except where otherwise noted, this item's license is described as
Αναφορά Δημιουργού-Μη Εμπορική Χρήση-Όχι Παράγωγα Έργα 3.0 Ελλάδα