Javascript must be enabled for the correct page display

Visualising large scale temporal geospatial multivariate graphs in a web-based environment

Groenewold, S (2015) Visualising large scale temporal geospatial multivariate graphs in a web-based environment. Master's Thesis / Essay, Computing Science.

master-thesis-stephan-groenewold.pdf - Published Version

Download (12MB) | Preview
[img] Text
toestemming.pdf - Other
Restricted to Backend only

Download (765kB)


There are many large scale infrastructures in our modern day society, such as gas transmission system, water supply networks and electrical grids. The behaviour of these kinds of networks can be described using simulations or sensors, using which huge datasets are created. Generally, we can refer to these types of datasets as large temporal geospatial multivariate graphs. The analysis of these datasets is important, since it can lead to new knowledge, for example by allowing researchers to get a better understanding of algorithms they are developing, or system monitors to detect problem areas. Analysing such datasets is however a challenging problem, due to their sheer size, as these datasets contain many attributes, describe large time ranges, and large spatial areas. In particular in a web-based environment where processing power, storage and bandwidth are limited, additional challenges are introduced. This work proposes an analysis environment where these types of datasets can be explored and analysed by using a set of linked visualisations. Each of the visualisations specialises in an aspect of the data, allowing them to complement each other. To serve the datasets to a web environment an aggregation and storage scheme is constructed, which attempts to give a good balance between enough detail and low bandwidth requirements. This functionality is added to an existing universal analysis framework. The result is a general purpose analysis platform for large temporal geospatial multivariate graphs. The implemented visualisations allow users to analyse aspects of their datasets which could previously not be viewed using their tools. In particular with respect to the analysis of temporal data significant improvements have been made. The solution combines a map based visualisation with an evolution spectrograph visualisation, which is a type of visualisation that to our knowledge has never before been used to display these types of graphs. Together these visualisation give the user the option to easily relate data back to the real world and view the structure of the network, while also offering a detailed view of the temporal behaviours in the data. The solution follows Shneiderman's mantra by giving a high level overview of the data, and allowing the user to zoom and filter the data, and view details as desired. We show such an analysis tool offers new non-trivial insights in the temporal aspect of these types of datasets.

Item Type: Thesis (Master's Thesis / Essay)
Degree programme: Computing Science
Thesis type: Master's Thesis / Essay
Language: English
Date Deposited: 15 Feb 2018 08:09
Last Modified: 15 Feb 2018 08:09

Actions (login required)

View Item View Item