Information Visualization: Principles, Methods, and Practice

This book was written to scratch my own itch: the book I wanted to teach out of for my graduate infovis course did not exist. The itch grew through the years of teaching my own course at UBC seven times, co-teaching a course at Stanford in 2001, and helping with the design of an early visualization course at Stanford in 1996 as a teaching assistant. I was dissatisfied with teaching primarily from original research papers. While it is very useful for graduate students to learn to read papers, what was missing was a synthesis view and a framework to guide thinking. The principles and methods that I intended a particular paper to illustrate were often only indirectly alluded to in the paper itself. Even after assigning many papers or book chapters as preparatory reading before each lecture, I was frustrated by the many major gaps in the ideas discussed. Moreover, the reading load was so heavy that it was impossible to fit in any design exercises along the way, so the students only gained direct experience as designers in a single monolithic final project. I was also dissatisfied with the lecture structure of my own course because of a problem shared by nearly every other course in the field: an incoherent approach to crosscutting the subject matter. Courses that lurch from one set of crosscuts to another are intellectually unsatisfying in that they make visualization seem like a grab-bag of assorted topics rather than a field with a unifying theoretical framework. There are at least five major ways to crosscut visualization material. One is by the field from which we draw techniques: cognitive science for perception and color, human-computer interaction for user studies and user-centered design, computer graphics for rendering, and so on. Another is by the problem domain addressed: biology vs. software engineering vs. computer networking vs. medicine vs. social/casual use. Yet another is by the families of techniques: focus+context vs. overview/detail vs. volume rendering vs. statistical graphics. Finally, evaluation is an important and central topic that should be interwoven throughout, but it did not fit into the standard pipelines and models. It was typically relegated to a single lecture, usually near the end, xi i i i i i i i i xii Preface so that it felt like an afterthought. This book is structured around a new model of visualization that unifies design and evaluation issues in …