Registered: 1465324029 Posts: 25
Reply with quote #1
On a recent call, I was presented with the following graphs. I have never seen these before and won't comment if they are good or bad, but they were certainly a bit confusing and challenging to interpret and make sense of.
Each of these "peaks" represent a category of products with several dots that each represent a single product. The more products in a category, the higher the peak. The second color layer tries to represent a relationship between the peaks. Not sure if this is arbitrary or not, but its hard to make sense of. It'd be interesting to get your thoughts on how to make this better. If there was a clear link between these categories, what better graphs could we represent that with. A simple bar chart can explain the peaks themselves for sure. Animated gif at the end!
Registered: 1135986598 Posts: 852
Reply with quote #2
Pseudo-3-D effects don't even work well for geographical displays, so it definitely isn't a good choice to use them for non-geographical information. People are attracted to the geographical metaphor, however, which is why we see attempts to use cartographic design practices for non-geographical displays from time to time. None of them work effectively. If there are meaingful links between product categories that must be dispayed, several options exist for doing this, depending on the nature of those relationships. For example, if the relationships involve a higher-level grouping of products (e.g., product lines), then you could use a bar graph as you suggested and color code the bars with a different color for each of those groups. If, on the other hand, the relationship between product categories involves some sort of interation between them, such as a cross-selling relationship, then a network diagram could be used, usually of the node-link variety. Imagine nodes represented as bubbles of varying sizes, in this case a bubble for each product category with the size of the bubble representing the number of products. The bubbles could also be color-coded to display a second variable, such as higher-level product groups or even a quantitative variable such as revenues. Now imagine lines linking the bubbles to represent relationships between the product categories, such as cross-selling relationships. The lines could vary in thickness to represent the strength of those relationships. It's difficult to discuss better solutions to this iwithout knowing for sure what they're trying to communicate. It's easy, however, to say that the approach that they used doesn't work. __________________ Stephen Few