A Visual Analysis of Ranking Predictions

MPhil Thesis Defence


Title: "A Visual Analysis of Ranking Predictions"

By

Mr. Abishek PURI


Abstract

Rankings are a natural and ubiquitous way of making decisions. In fact, the 
very definition of making a choice requires you to rank all options and choose 
the best one. However, as the number of options grows, we are unable to rank 
all available options individually, and so outsource this task to ranking 
systems. One such outsourced task is the ranking of universities for studying. 
As it is infeasible for a student to visit all universities and rank them 
individually, they often refer to published rankings such as QS instead. A 
large body of work exists that shows how the students can be helped to make a 
better decision, but there is another set of users who are affected by such 
systems: the universities being ranked. In this case, the universities 
themselves have a vested interest in understanding how the data they submit and 
the actions they take will affect their ranking position in the future. This is 
a critical issue for a university because any change in its ranking position, 
no matter how slight, can affect its access to high quality students and 
external funding. This is a challenging problem, as the analysis must allow the 
universities to visualise the entire space of scenarios for the following year, 
depending on how the submitted data is changed this year. It must also allow 
the universities to compare their predictions to the expected movement of rival 
universities, to see if they are outperforming their rivals in key areas as 
well as overall. In this thesis we present RankBooster, a visual analytics 
system that explores the set of scenarios for the universities future ranking 
position and allows the university to compare itself to its rivals. We use 
multiple case studies conducted with university ranking experts to evaluate the 
effectiveness of our system. We also present a novel abstraction of our given 
users tasks, as to the best of our knowledge we are the first researchers to 
tackle rankings visualisation from this perspective.


Date:			Monday, 10 June 2019

Time:			3:00pm - 5:00pm

Venue:			Room 3494
 			Lifts 25/26

Committee Members:	Prof. Huamin Qu (Supervisor)
 			Dr. Pedro Sander (Chairperson)
 			Prof. Mordecai Golin


**** ALL are Welcome ****