Eugène Horber is Professor of Applied Computer Science in the department of political science and international relations at the University of Geneva. He teaches social science methodology, applied computer science, and statistics to social scientists. His research interests and publications are in the area of statistical methodology, aggregate data analysis, data visualization and applied computer science (didactical software, hypertext), as well as computer assisted qualitative analysis. He is the author of a software package for exploratory data analysis.
This course introduces the principles and tools of visual data analysis. Its main aim is to show that visual tools are as essential to statistical practice as are numerical tools. Although graphical tools have always been used by statisticians and data analysts, it is only with the publication of John Tukey's pioneering 1977 book "Exploratory Data Analysis" that visualization became far more concrete and effective, as well as an important field in statistical research. Modern (statistical) software offers a great many graphical and visual tools, often ignored by both users and teachers. In practice graphical tools are still often limited to barely more than flashy barcharts and some scatterplots and not considered "serious" statistical tools, because no mathematical formulae are shown.
Visual, exploratory tools have recently become an essential component of Data Mining, Knowledge discovery, Visual analytics or similarly labelled areas, stressing data and goal driven analysis, where visual interaction with the user plays a key role.The course will introduce a wide variety of tools showing how data can be analysed visually. Its main aim is to stimulate a general attitude and awareness towards methodological problems based on a creative visual approach. Special attention will be paid to integrating visual statistics into current practice using standard software.
The course is based on the following methodological and didactic principles:
The workshop will have several strands. A linear strand presenting tools for univariate analysis and comparison, then visual tools for studying relationships between two and more variables and finally multidimensional tools. Other central themes will underpin the first, namely the philosophy of visual data exploration (as a paradigm for statistical analysis), visual communication (perception and design), ergonomic issues (tool design, software), as well as integrating visual tools into your current practice.
Except some basic skills with statistics and statistical software there are no specific prerequisites for this workshop except motivation. You should be open minded and flexible, do not expect streamlined text-book style solutions, but be prepared to adapt and integrate the various tools presented to particular data and research problems and critically review your methodological practice. The course should appeal to both beginners (introduction to statistics the visual way) and more advanced users (providing an alternative, complementary way of looking at statistical problems).
You are encouraged to use the tools and techniques presented in the statistical software your are most familiar with. Ample on-line information will be provided for SPSS, Stata and R to assist you in the practical exercises and group projects.