Instructor: Guy Melançon (email:
Guy dot Melancon at labri dot fr)
Information Visualization and Visual Analytics are hot topics. Lots of information, techniques, software librairies, research results can be easily found on the web.
When you think about it, it seems science is not done like it used to (Gray 2007).
Science as it was practiced during the last millennium has tried to describe phenomenon of the physical world by mastering techniques necessary for their explanation or reproduction: that science was fundamentally empirical (think of Galileo).
A few centuries later, science gained predictive power in developing models and formulating laws. Theoretical tools allowed postulating theories that could then be confirmed or refuted by observation (think of Newton and the universal law of motion).
The last decades have seen an increase in the complexity of the models formulated to explain phenomena. The analytical approach has given way to the simulation of models requiring intense calculations: science has accelerated its development by becoming computational. (The recent performance of computers playing chess or Go against world expert is a good illustration).
These computing resources today allow the production of increasing volumes of abstract data with or without intrinsic geometry. We agree with (Gray, 2007) and enter the arena of this new Science of data exploration. The leitmotiv is that unsuspected phenomenon at play may be identified by exploring data, hidden structures may be uncovered to explain complex phenomenon.
Data processing and mining are nowadays a priority both in research and industry, small or large. Organizations now produces loads of data on their internal processes as well as from their clients. Technological or strategic watch requires to collect and forage all available data in order to gain competitive edge over concurrent organizations, or to better position themselves in the socio-economical context.
Data can be massive – as in “big”. Data is also complex, just as the phenomenon they emerge from, but also because it often is non-structured. Seeking for the information encapsulated within the data is difficult, discovering new insights from data is challenging.
Data mining develops algorithms capable of identifying structural patterns, expressed as association rules for instance. A classical example is that of a supermarket seeking at better knowing consumers’ habits, typically through the use of fidelity cards or mobile application account, in order to refine its marketing strategy. A telecom company will foresee the possibility to better understand how the teenager market differs from adult consumers. Data mining results can be complex just as the data that needs to be processed ; they often need to be sorted, organized or classified before decision can be made.
Data mining is all about discovering structural patterns. Information visualisation relies on a crucial observation, that around 40% of our cortex activity is devoted to processing visual signals.
This is a rather common way of looking at things shared by the InfoVis community. The good thing about it is that everybody agrees on it. Another good side of it is that the pipeline tends to conciliate both the user's perspective and the engineering perspective -how you build a viz system). The bad side of it is it focused on the automated side of InfoVis (processes and algorithmic) rather than on cognitive processes.
Visual Analytics is (as defined by Cook and Thomas 2006) the science of analytical reasoning facilitated by interactive visual interfaces. (See the InfoVis Wiki)