Data Visualization: Telling Stories At A Glance
Thousands years ago, human drew figures on rocks, inside their caves, on vases . Centuries passed, those figures was simplified, alphabet was found and replacing those complex pictures, and now we use text as an easier, most common way to communicate non-verbally. Knowing this, can we say comprehending pictures or shape is actually a basic human instinct?
So if we can communicate with pictures, correctly and effectively, it should be better than using text right? Since text is actually harder to comprehend because to understand it we have to recognize the shape of letters, the combination of shapes to form word, recognize the relation between words, and finally comprehend the actual meaning of the string of words.
Here’s a data about people’s hobby:
Which one is easier, faster one to get information about the most popular hobby? Isn’t it easier to find the information using the chart? If not then you might want to stop reading altogether since I failed to convince you to use visualizations.
I think the most important thing about data visualization is to communicate the stories in the shortest way possible. The best visualization is when people can get at least a piece of information with only a glance.
But how?
As visualization is actually a form of communication and there’s something called visual grammar. Knowing this grammar, or visual language, could really help you on deciding how to present or visualize your data.
The most basic visual language elements are color, shape, size and position.
I think just by actively consider these elements in our mind, we can make a good data visualization.
So there are several things we may want to communicate with data visualizations and I’ll try to explain how to make a visualization to communicate that based on the visual grammar.
How much…?
— or magnitude, the different values between entities. When we want to tell the story about magnitude (and only magnitude), we use visual language: size. And the easiest way to use the size language is by using a bar chart.
For example, to communicate how much is the profit of these four restaurant:
With this chart you can see how much profit of each restaurant and you can easily compare which has the most and least profit within a glance.
What percentage is…?
— or to know how a single group broken into parts. Since percentage is also a value, we’ll also use visual language: size. But every element is a part of a whole group so we need to show that it’s a part of a same group but also a different element. To show that every element is a part of the same group we’ll use shape (same shape)and to show that each element is a different element we’ll use colors (different color).
Although it sounds complicated, one of the charts which can represent both 3 visual language is pretty common and easy: pie chart. For example to communicate the percentage of women in a company:
Within a glance, we can see the percentage of each gender. The impact is bigger than just seeing the numbers. You can actually see the proportion of each gender.
You might notice that the round shape shows that it’s a single group: employees. The different colors show that it’s a different element: male and female. And the size of each colors shows the value of the percentage. Easy, huh?
How’s the trend…?
— or how’s the value over some period of time. So now you might know what to do already. Visual language to represent value? Size. To show time, we’ll use position. So for example, we want to show trend of average temperature everyday through July.
Not bad! We can see that the temperature goes slightly higher on 16 July. But is it only slightly? The lowest temperature is on 3 July, 35 °C and the highest is 47–48 °C in 16 July. The difference is not slightly at all! Even an air conditioner usually has only have 10 degrees range. But how to communicate that using a chart?
Now, let’s say the normal temperature in July is above 33 °C so we can actually ignore the 33 degree part. But bar represents a relative size and we can’t just cut below the 33 degree because it’ll mess the relative size. So since size doesn’t really work here, let’s use visual language position instead of size.
While this new chart shows that the temperature has spike on 16 July, it doesn’t look make sense because each dots seems like a different entity than the rest. So how to make it look like they represent the same thing (temperature)? We’ll connect the dots to show that it’s a same entity just in a different position through the period.
Much better! Now we can clearly see (in a glance!) that the temperature goes super high in the middle of July and goes back down to the late July.
Turns out making data visualization is not that simple. It has multiple ways to do and everything is really depends on what you want to communicate to your audience.
Next time, more about selecting chart for your data visualization. Until then, farewell.
