Saw this great little sign at Ippudo, the ramen shop, the other day:
It's a great example of highly effective data visualization. The names on the board are sake brands.
The menu (a version of a data table) is the conventional way of displaying this information.
Customers are selecting a sake. They don't have a favorite, or don't recognize many of these brands. They know a bit about their preferences: I like full-bodied, or I want the dry one.
On a menu, the key data are missing. So the first order of business is to find data on full- and light-bodied, and dry and sweet. The pricing data are omitted, possibly because it clutters up the design, or because the shop doesn't want customers to focus on price - or both.
The design uses a scatter plot. The customer finds the right quartet, thus narrowing the choices to three or four brands. Then, the positions on the two axes allow the customer to drill down further.
This user experience is leaps and bounds above scanning a list of names, and asking someone who may or may not be an expert.
Back to the Data
The success of the design depends crucially on selecting the right data. Baked into the scatter plot is the assumption that the designer knows the two factors most influential to the customer's decision. Technically, this is a "variable selection" problem: of all factors determining the brand choice, which two are the most important?
Think about the downside of selecting the wrong factors. Then, the scatter plot makes it harder to choose the sake compared to the menu.