What is Visual Encoding?
In an imaginary world, where the human eye would be perfect, we could even see atoms side by side. And imaginary speaking, concepts like color, angle, size, height, position, shape would have been probably unknown. In our world things are quite different, and as far as it concerns our visual perception, all the mentioned qualities comprise our basic tools to interpret it.
What about the world of data? Is it closer to the imaginary or the real world? I would choose the first, as in my mind, raw data is just strings and number characters side by side. I just mentioned the phrase in my mind and that is because what really matters, after all, is how human perception can understand data in a more meaningful, direct, and intuitive manner. To succeed that, visual objects must be used. And here comes visual encoding, which is the way data are represented in visual objects. In the following picture, we can see which are the ways we can encode our data.
Visual Encoding Options
So how can I bring the world of data closer to mine?
The answer is simple: Just visualize them!
But to do it right we should first understand our data.
A way of decoding our data is to categorize them according to their level of measurement.
Data can be either qualitative or quantitative.
Qualitative data consist of attributes, labels, and other non-numerical entries. Sometimes this is called “categorical” data.
Quantitative data consist of numerical measurements or counts.
Qualitative and quantitative data can be classified into four different levels of measurement: nominal, ordinal, interval, and ratio.
Why categorizing our data according to their level of measurement is helpful?
There are some basic rules about which is the most accurate visual encoding, according to the level of measurement our data belongs to. Position is the best option for all kinds of data either they are qualitative or quantitative.
But what if we want to use color? Would, using color saturation when visualizing Nominal Data be helpful? Does it work the same way when visualizing Ordinal data? Will the use of volume give a more direct message than length? When is it meaningful to use different symbols like squares, triangles, and circles for representing our data?
All these questions can be easily answered by the following guide, which gives the most to the less accurate visual encoding, according to our data.
But then more questions come. Which is the best way to compare values? Does position work in comparisons? Which is the most aptly way to show distribution? What if I want to show trends over time?
Supplementarily to the first guide, another guide of classifying visualizations according to the question: What do we want to communicate, can lead us to the correct data visualization choice.
What do we want to communicate?
We use data visualizations to show Comparison, Composition, Distribution and Relationship. The following diagram summarizes the most suitable charts per case.
Available Qlik Sense charts per case
Visualizations used for Comparison, highlights the size of our data, comparing each other and it is the best way to discover the highest and the lowest values, or compare current values versus old values. These visualizations can be used to answer questions such as: Which are the top sellers of our organization, how are our number of orders compared to last year or how are our sales trends formed for different regions, over time.
Best charts for comparison in Qlik Sense are:
Composition schematizes which is the share of an individual part compared to the total. They can be used to show us either relatives or absolute values of the whole. Questions that can be answered using these kinds of data are: What is the sales share per salesperson, how margin proceeds from gross sales, or which is the monthly production share per department.
Best charts for comparison in Qlik Sense are:
Distribution charts help us discover the shape of our data, meaning, the range of values, the prevailing tendency, and the outliers. Questions that can be answered analyzing distribution can be: What is the frequency of prices among products, Which of our customers are outliers, or Which is our sales distribution over products per product category.
Best charts for showing distribution in Qlik Sense are:
Visualizations used for discovering relationships in our data can show us if one variable affects another variable and in which manner. They can lead us to identify patterns, clusters, correlations, and outliers in our data. We can answer questions such as: How discount impacts on our customers sales, what is the correlation between customers rating and number of orders or if advertising cost per month has a negative or positive impact on our sales.
The native Qlik Sense chart to discover relationships is: