I currently teach a workshop called “Information Visualization for Discovery and Analysis”. In this workshop I present several general practices for analyzing quantitative business data using interactive visual techniques, as well as many specific visual representations of data and visual analysis techniques that are designed to solve particular business analysis problems. For instance, I describe several ways to display and interact with time-series data to solve particular problems, including one that I’ll describe here as a example. My purpose here is to illustrate the kind of examples that I plan to include in my next book and to invite you to share your own visual analysis designs and techniques for making sense of quantitative business data. If you submit a technique that is new to me and I decide to include it in my book, I will gladly give you the credit for passing on the idea and will send you a complimentary copy of the book once it’s published. Much of what I know I’ve learned from people like you who work regularly in the trenches to solve real-world business problems. Please submit your ideas via the Discussion Board (see the discussion forum on “Visual data analysis techniques”) on my web site at www.PerceptualEdge.com.
On occasion it is useful to compare rates of change between two or more sets of time-series data. For instance, you might want to determine if your domestic or international sales are increasing the fastest. Line graphs do a great job of showing the ups,downs, and overall trends of values as they change through time. Here an example of how your domestic and international sales might look during the last six months.
It is natural when looking at a time-series graph such as this to assume that the orange line is increasing at a faster rate than the green line, but in fact that are increasing at precisely the same rate. A 10% increase beginning at $10,000 equals $1,000, while a 10% increase beginning at $100,000 equals $10,000, and on a standard quantitative scale the slope of a line that increases by $1,000 is much less than one that increases by $10,000. This does not hold true, however, for logarithmic scales. The same data appears in the graph below, this time using a logarithmic scale. Equal rates of change when using a logarithmic scale appear as equal slopes, no matter how much the actual values are or the differences between them.
Therefore, if you want to compare rates of change rather than actual amounts of change, using a logarithmic scale makes the comparison easy.