Datavisualization as human-data mediation

While datavisualization has long been a field of scientific research expertise, particularly in the mathematical, statistical and computer sciences, it has also been built on interdisciplinarity with the humanities and social sciences (SHS): cartography, art and design sciences, and cognitive sciences have all contributed to its development. While it is difficult to draw up an exhaustive history of datavisualization, the socio-historical foundations of graphic representation practices emerged very early in history. As early as the 2nd century, with cartography, emblematic of the need to represent reality in order to act on it in return. In the 18th century, with the rise of mathematics and its statistical side, the graphic representation of data underwent a decisive turning point with the appearance of the first graphs in William Playfair's Commercial and Political Atlas (1789), which are still widely used today: evolution curves, bar graphs and pie charts. But it wasn't until the 1970s that data mining was fully popularized. Francis Anscombe, a statistician, demonstrated with a series of four data sets, the famous "Anscombe Quartet", that data presented in tabular form are not easy to understand. These four sets have simple and fairly similar statistical properties in their linear tabular representation (mean, variance, correlation and regression have close values). However, when represented in graphical form, the differences between the four data sets can be seen, demonstrating the value of this representation.

The democratization of datavisualization spread beyond the scientific world, notably with the work of John Tukey, professor of statistics at Princeton University and author of Exploratory data Analysis, a book on data analysis and presentation methods. Subsequently, from the 1980s onwards, designers David McCandeless (2012), Stephen Few (2006) and Manuel Lima (2011) gave this discipline its full aesthetic dimension.

Today, in the context of Big Data, datavisualization is a powerful tool for mediating between man and data, enabling us to reason from data to grasp the complexity of the world. As such, it represents a major innovation challenge for both scientific research and organizations. Its applications can be found in many fields, such as statistical and decision-support computing for organizations, data science for biology and genomics, the humanities and social sciences with digital humanities, digital cartography and visual network analysis.