Data Science

Until now, it is certain that no one could have imagined the establishment of such a "science of data." However, just as physics studies the motion, structure, and interactions of matter, and chemistry studies the changes, components, and properties of matter, "data science"—the study of the structure, effects, components, and properties of data—is now a respectable scientific field in its own right.

Of course, such a science did not emerge suddenly. Driven by the need to build and operate large-scale databases, research into relational databases began in the 1970s, and research into data mining to discover hidden laws in large-scale data also flourished. Statistics, the traditional academic discipline for dealing with data, also serves as its foundation.

The history of statistics is long, dating back to the 17th century. Spurred by the great plague, it developed in the form of demographics out of the necessity to calculate mortality rates. It also evolved as a form of statistics known as *Kokujogaku* (the study of national conditions), based on the view that understanding the concrete state of nations was as important for statecraft as understanding the human body was for medicine. Yukichi Fukuzawa was also a pioneer who pointed out and practiced the importance of data-based logical reasoning, such as by publishing *Bankoku Seihyo* (Tables of the World) and discussing the relationship between marriage rates and rice prices in "An Outline of a Theory of Civilization."

While statistics, originally the science of data, seemed to become merely a tool for developing formal logic as it transformed into modern statistics based on probability theory, it is now regaining its brilliance within the framework of data science. By treating data as an abstract object of study, data science has become a general-purpose field of research not tied to any specific domain. Widely applicable in modern society, which must handle large and complex data, data science continues to make significant progress, giving rise to high-dimensional data visualization technologies represented by its infrastructure, such as DandD (Figure 1) and Textile Plot (Figure 2).