This article is part of an Essential Guide, our editor-selected collection of our best articles, videos and other content on this topic. Explore more in this guide:
2. - Data visualization tools evolve with big data: Read more in this section
- Achieving big data analytics success with data visualization tools
- Data visualization tools offer more capabilities -- and challenges
- Users feel sense of ownership when they can visualize big data
Explore other sections in this guide:
- 1. - Flash vs. function in dashboard design and development
- 3. - News and trends in dashboards and data visualization
- 4. - Dashboards and data visualization terms, defined
Demand for data visualization tools is rising sharply, partly as a result of more companies seeking to gain valuable business insights through “big data” analytics initiatives. But achieving success with data visualization often requires fresh thinking about how to present information to business users, especially in big-data environments, according to data management analysts.
Data visualization -- which enables users to create graphical and often interactive representations of data sets big and small -- can contribute greatly to improvements in corporate business intelligence (BI) efforts and organizational productivity, based on the results of a survey conducted in 2010 by The Data Warehousing Institute (TDWI).
We're seeing a very positive association between advanced visualization and analytics. People just assume the two go together.
Phillip Russom, director of data management research, TDWI
In the online survey of 210 BI practitioners and business users, 74% of the respondents said the influence of data visualization on attaining business insights within their organizations was “very high” or “high,” while another 23% said the technology had a “moderate” influence on the BI process.
The survey also indicated that the use of data visualization tools was helping to drive increased adoption of BI dashboards, typically the preferred medium for accessing and viewing charts, maps, graphs and other visuals. And more recent TDWI studies point to a growing confluence between data visualization software and big-data analytics.
“My own research shows that advanced data visualization is the No. 1 area where users expect to buy more tools, and use more, as a best practice with big-data analytics,” said Philip Russom, director of data management research at TDWI. “We're seeing a very positive association between advanced visualization and analytics. People just assume the two go together.”
Effective data visualization is key to obtaining value from a big-data analytics investment, agreed Stephen McDaniel, co-founder of Freakalytics LLC, a Seattle-based visual analytics consulting firm. But McDaniel, who spoke at a recent TDWI forum on big-data analytics in Orlando, Fla., said there are some practical limitations that organizations need to consider before jumping into the process of visualizing big-data analytics findings.
Finding the right big-data mix for data visualization tools
For example, attempting to populate graphs with, say, billions of individual data points might not be an effective way to get started. “The typical monitor that is sitting on your desk can only show 200,000 points,” he noted. “If you get one of these advanced monitors, let's say a few million points is possible. But that’s still nowhere near billions of records.”
Organizations looking to visualize big data can overcome such limitations by aggregating data or eliminating unnecessary data points based on the specific requirements of the intended user audience, McDaniel said.
Effective chart selection and good dashboard design is also important. “You want to have well-designed management dashboards,” McDaniel said, adding that even if you can only “put a small amount of data on the display,” business executives and other end users should be able to drill down into the underlying information to conduct ad hoc queries and find the answers they’re looking for.
In addition, Russom recommended that when creating visual representations of big-data analytics results, think twice before overlooking or eliminating outliers -- data points about, for example, customers whose spending behavior lies far outside the norm. He added that paying attention to such bits of information can help organizations with tasks such as detecting fraud and identifying new customer segments -- big spenders, for example.
“If you go scrubbing and cleansing that data, you’ll lose some of those funky little nuggets that you’re looking for,” Russom said -- and as a result, he noted, you could “lose value from an analytic viewpoint.”
Despite its growing popularity, data visualization technology has some general drawbacks to take into account, according to McDaniel. For one thing, he said, most people have never been taught how to properly use data and graphics to “tell a story.” The result, he added, is that they often create charts, graphs and other visuals that are difficult to understand and fail to get the intended point across.
Data visualization problem: Too much information
McDaniel said the best way to overcome that problem is to focus on the questions that need to be answered and keep the visuals as simple and clear as possible. “If you can't present data and explain it to your business sponsors and your business users, you'll really have a very hard time succeeding,” he said. “It doesn't matter how great your ideas are.”
Another common mistake that users of data visualization tools make is relying on the wrong kinds of graphs to tell a story. For example, McDaniel said “a lot of studies” have shown that bar charts get information across more effectively than pie charts, partly because they can display far more data points without becoming difficult to understand. “I've made a bar chart where you can see 50 bars and actually interpret them,” he noted.
Another mistake cited by McDaniel is overkill -- using unnecessary 3-D graphics, pop-ups, animations or other frills and completely overshadowing the point of the visual representation.
“The reason visual analytics is so important is the iterative questioning of the data that it enables,” he said. “And unfortunately, while 3-D graphs look cool and interesting and people love them, there’s a lot of evidence [that] we don’t process them or understand them well.”
One of the best ways to ensure that visual representations of data are effective is through increased communication between business and IT, added Elaine McDaniel, a data visualization consultant and Freakalytics co-founder.
“Ask people's opinions. Get feedback. Incorporate their opinions and get people working together,” she said. “From all of the [business analysts] we've dealt with over the years, the biggest complaint is that they cannot communicate with people in IT.”
Conference attendee James MacGillivray, a BI architect with a company in the agriculture industry, agreed that it’s important to spend time getting the right requirements from business users before moving forward with a data visualization -- or big-data analytics -- project.
“Too often, we are trying to build it and think that they will come. That’s the wrong approach,” said MacGillivray, who asked that his company not be identified. “You should have the business user drive [the project], because I don't want to drive it. I'll facilitate it.”