In my final year of undergraduate, my friend and I, both being very passionate football players, began working on a project to visualize some data about football players. We both threw different ideas at each other, and even though we ended up going in different with our projects, I gained significantly from our exchanges. Two years later, and much wiser, I have worked on 5 projects in Infovis so far, which includes both research work and course projects. I have collated all my work in Infovis here:
I am collaborating on a research project with Prof. Matthew Kay to study how different ways of presenting the uncertainty around a statistical estimate can affect how the viewer interprets the data and can make the uncertainty in that estimate easier to understand.
The main motivation of this study was that researchers often tend to misunderstand the uncertainty in statistical results; for example, they have been shown to be susceptible to several statistical fallacies when interpreting the meaning of 95% confidence intervals. These misunderstandings are prevalent even amongst those who have some education in statistical methods.
Methodology: We are showing alternative visualizations of uncertainty to researchers in the context of a hypothetical study result, and then testing their interpretation of the uncertainty using questions adapted from previous studies of statistical understanding. We are testing whether changing the type of estimate shown - a Bayesian or frequentist estimate - can elicit more correct interpretations of the results. This work has the potential to influence recommendations for statistical methods and reporting across scientific fields that make use of confidence intervals.
I began this project due to my strong interest in Football and my curiosity to understand and analyse player transfers. Due to constraints of collecting data, I focused on the top 5 clubs in the Premier League and gathered data of each club's transfer activity.
This data included players bought, sold or released, player performance data (appearances and goals) and team event data such as trophies won, managers changed, ownership change etc. The goal was to identify the factors which determine the success of a club in the long run. I analysed the player transfer fees and their corresponding impact at the club to find a correlation.
The group project allowed us to explore a dataset and create a visualization which affords different types of tasks based on the role of an user that we identified. The project is to investigate how Americans engage in different artformss (such as opera, jazz etc.) using the Survey of Public Participation in the Arts 1982-2012 (ICPSR 35596).
We created a visualization which allowed the user understand the relationships between demographic variables and art forms, as well as the relative popularity of the different artforms.
The individual project explored the topic of Explorable Explorations - the use of interactive visualizations to support a learning task. Here, I explored how this can be used to better understand mathematical paradoxes such as the St. Petersburg Paradox.
I was inspired to pursue this due to my own fascination with the Monty Hall paradox and Paul Erdos's disbelief in the correct solution
Why? To encourage active reading, several ideas have been proposed: A reactive document allows the reader to play with the author's assumptions and analyses, and see the consquences. An explorable example makes the abstract concrete, and allows the reader to develop an intuition for how a system works. Contextual information allows the reader to learn related material just-in-time, and cross-check the author's claims. (Bret Victor; Explorable Explations)