Visualize This: The FlowingData Guide to Design, Visualization, and Statistics

What to Look For

It’s easy to compare across a single variable. One house has more square feet than another house, or one cat weighs more than another cat. Across two variables, it is a little more difficult, but it’s still doable. The first house has more square feet, but the second house has more bathrooms. The first cat weighs more and has short hair, whereas the second cat weighs less and has long hair.

What if you have one hundred houses or one hundred cats to classify? What if you have more variables for each house, such as number of bedrooms, backyard size, and housing association fees? You end up with the number of units times the number of variables. Okay, now it is more tricky, and this is what we focus on.

Perhaps your data has a number of variables, but you want to classify or group units (for example, people or places) into categories and find the outliers or standouts. You want to look at each variable for differences, but you also want to see differences across all variables. Two basketball players could have completely different scoring averages, but they could be almost identical in rebounds, steals, and minutes played per game. You need to find differences but not forget the similarities and relationships, just like, oh yes, the sports commentators.