No Cover Image

E-Thesis 83 views 153 downloads

Visualisation of Large-Scale Call-Centre Data / DYLAN REES

Swansea University Author: DYLAN REES

  • Rees_Dylan_G_PhD_Thesis_Final_Redacted.pdf

    PDF | Redacted version - open access

    Copyright: The author, Dylan Geraint Rees, 2020.

    Download (42.62MB)

DOI (Published version): 10.23889/SUthesis.56839

Abstract

The contact centre industry employs 4% of the entire United King-dom and United States’ working population and generates gigabytes of operational data that require analysis, to provide insight and to improve efficiency. This thesis is the result of a collaboration with QPC Limited who provide data c...

Full description

Published: Swansea 2020
Institution: Swansea University
Degree level: Doctoral
Degree name: Ph.D
Supervisor: Laramee, Robert S.
URI: https://cronfa.swan.ac.uk/Record/cronfa56839
Tags: Add Tag
No Tags, Be the first to tag this record!
Abstract: The contact centre industry employs 4% of the entire United King-dom and United States’ working population and generates gigabytes of operational data that require analysis, to provide insight and to improve efficiency. This thesis is the result of a collaboration with QPC Limited who provide data collection and analysis products for call centres. They provided a large data-set featuring almost 5 million calls to be analysed. This thesis utilises novel visualisation techniques to create tools for the exploration of the large, complex call centre data-set and to facilitate unique observations into the data.A survey of information visualisation books is presented, provid-ing a thorough background of the field. Following this, a feature-rich application that visualises large call centre data sets using scatterplots that support millions of points is presented. The application utilises both the CPU and GPU acceleration for processing and filtering and is exhibited with millions of call events.This is expanded upon with the use of glyphs to depict agent behaviour in a call centre. A technique is developed to cluster over-lapping glyphs into a single parent glyph dependant on zoom level and a customizable distance metric. This hierarchical glyph repre-sents the mean value of all child agent glyphs, removing overlap and reducing visual clutter. A novel technique for visualising individually tailored glyphs using a Graphics Processing Unit is also presented, and demonstrated rendering over 100,000 glyphs at interactive frame rates. An open-source code example is provided for reproducibility.Finally, a novel interaction and layout method is introduced for improving the scalability of chord diagrams to visualise call transfers. An exploration of sketch-based methods for showing multiple links and direction is made, and a sketch-based brushing technique for filtering is proposed. Feedback from domain experts in the call centre industry is reported for all applications developed.
Item Description: ORCiD identifier https://orcid.org/0000-0002-7129-5825;Redacted: A selection of third party content is redacted or is partially redacted from this thesis.
College: College of Science