What is Data Visualization?
Content Marketing Manager, Sigma
Half of our brains are directly or indirectly devoted to processing visual information. A team of neuroscientists from MIT discovered that the human brain could process entire images the eye sees for as little as 13 milliseconds. And with 2.5 quintillion bytes of data being produced every day, visualization has emerged as an effective way of using data to tell a story, learn new insights, and influence business decisions.
Let’s dive into what data visualization is, as well as how to use it effectively to answer critical business questions.
Data visualization defined
Data visualization describes any effort to help people understand the significance of data by placing it in a visual context. We use visualization to quickly make sense of data, which would otherwise be difficult to draw conclusions from. Its main objective is to distill large datasets into visual graphics to allow for an easy understanding of complex relationships within the data.
Rapid advancements in data visualization software allow us to quickly identify patterns, trends, and correlations that often go undetected in text-based data. Many data visualization tools also come with dashboard functionality that enables users to track and monitor multiple key performance indicators and business metrics in one place.
Take a deep dive into data visualization with our definitive online guide.
Identifying your data question
All data problems start with a question. Why? Answers are meaningless if you don’t understand the question.
According to Microsoft’s Senior Data Scientist, Brandon Rohrer, there are only five questions that data science can answer.
Is this A or B? – Useful for any question that has just two possible answers. For example, “Which brings in more qualified sales leads – an eBook or a White Paper?”
Is this weird? – This is all about detecting anomalies, which provides clues about where to look for problems. Think when your email flags a specific message as spam, perhaps because it was sent from a spoofed domain.
How much, or how many? – Here’s where data can help you make numerical predictions, such as, “Will revenue increase in Q2?”
How is this organized? – This helps us to understand the structure of a data set, which in turn makes it easier for us to interpret the data. A question you might ask is, “Which type of handset has more bugs in a specific version of the mobile app?”
What should I do next? – Reinforcement learning is based on algorithms that learn from outcomes, and decide on the next best action — usually by a machine or a robot. For instance, if using a chatbot on your website to greet visitors, the question might be, “If I’m asked X, should I refer them to the support documentation or connect them to a human rep?”
But you can’t just ask any question.
Imagine you went to visit a fortune teller to ask what your future holds. It’s in this person’s best interest to keep the answer as vague as possible as not to risk giving the wrong response. She may tell you, “You’re going to be happy, healthy, and rich.”
That sounds great, but it doesn’t tell you much about what to expect. Instead, if you throw out an uber-specific question like, “In what year will I be promoted to CMO?” or “What’s my future spouse’s name?”, you may actually get the kind of response you’re hoping for.
Likewise, you’ll want to ask an airtight question of your data to have any chance of receiving a useful answer.
Data visualization fundamentals
Before data visualizations can give you the answers you want, you have to follow a systematic process to obtain, organize, and analyze the data. Just like following a recipe, the more you stick to the steps, the more likely you’ll end up with a tasty dish.
Data visualization in 7 steps:
Acquire – First, you’ll need to obtain the data from your on-premise servers, third-party software applications, or cloud-based storage service.
Parse – Next, tag and categorize each part of the data by its intended use to give it more structure and meaning.
Filter – Remove any aspects of a data set that aren’t relevant.
Mine – Leverage statistics and data mining to uncover patterns or put the data in a mathematical context.
Represent – Decide what type of visualization will best represent your data.
Refine – How can you use design principles to make your data visualization easier for the viewer to understand?
Interact – Give the user the option to manipulate the data and control what features are visible to them.
If you choose to deviate from the recipe, there’s still a chance you’ll end up with something edible; however, this path comes with inherent risk. By sticking to the fundamentals of data visualization laid out by the top data scientists in the industry, you’re more likely to come out of it with a visualization that answers an important business question and can be used to gain deep insights that help drive the business forward.
Sigma is always adding new types of visualizations to our tool. To see which of these are currently supported, visit our help center.
Still have questions about data visualization? Read our definitive guide to learn more.