This event will take place in the Leverhulme Library (COL.6.15).
Title - Matrix Visualization for Exploratory Data Analysis.
Abstract - "It is important to understand what you CAN DO before you learn to measure how WELL you seem to have DONE it" (Exploratory Data Analysis: John Tukey, 1977). Modern statisticians and data scientists are confronted with the challenge of deciphering increasingly high-dimensional and complex data, as traditional graphics and visualization tools fall short in meeting these demands. The ability to comprehend the overarching structure in massive datasets is an even more formidable task. Therefore, effective Exploratory Data Analysis (EDA) approaches, coupled with intuitive and user-friendly data visualization environments, will become increasingly vital in determining what can be achieved in the era of big data.
Matrix Visualization (MV) has proven to be more effective than standard EDA tools, such as Boxplot, Scatterplot (utilizing dimension reduction techniques), and Parallel Coordinate Plot, in extracting information from moderate to large (dimension and sample size) datasets. In this presentation, I will initially provide a concise overview of MV's technical foundation for continuous, binary, nominal, symbolic data, as well as data with cartographic links, using the Generalized Association Plots (GAP) developed by our information visualization laboratory. Subsequently, real-world applications addressing scientific issues in fields such as biological experiments, medical research, and social surveys will be showcased, followed by a discussion of ongoing advancements and potential future trajectories for MV research.
Bio - Website