Information Visualization for Business – Past & Future

Information Management Magazine, January 2005

Richard Brath, Michael Peters

It has been almost ten years since the release of Windows 95. Hidden in Windows 95 was a library containing functionality never before seen on the desktop - Open GL. Open GL, a standards-based enabling technology, made creating interactive 3-D graphics faster, easier and more dependable than before.

Now, 10 years later, where have these 3-D graphics gone to provide business advantage to those working with data? Instead of the virtual reality interfaces postulated in 1992, interactive desktop 3-D has become much more effective at solving specific business needs in monitoring, analysis and reporting and is now broadly referred to as information visualization and visual analytics.

Business systems are generating ever-increasing amounts of data, and more and more of this data is moving in real time. Data mining is a standard feature in some databases. Automatic analytic modeling engines are found in many applications. However, there remain many applications where human judgment is required, including: management reporting, risk management, model exploration and validation, real-time operations and information security. With all this rich information, the challenge remains to derive the most business value from the ever-increasing amount of available information.

Visual information can significantly improve productivity. Users can explore large amounts of data, rapidly assimilate information from many sources, reason with it, understand it and create new knowledge based on it. With the right visual picture, people can make better decisions, faster, backed with more information.

The last ten years has witnessed an explosion of graphics in reporting and analysis software - dashboards, scorecarding and rich charting techniques - which have significantly enhanced information presentation. Interactive features, such as linking and drill-through, are becoming commonplace. Visual techniques using maps, gauges, stoplights, radar charts are now widely available for reporting applications. Techniques such as zooming and integrated correlations (where all charts are synchronized so that selecting one bar in a chart updates all active charts to highlight the selected subset) are becoming more widespread.

This first generation of visualization has familiarized the corporate world with using and interacting with business graphics. Now, some corporations are looking to harness visualization for strategic advantages in their core competencies, adding advanced visualization to key monitoring, analysis and client-facing processes.

Monitoring

One of the most obvious benefits of visualization is helping people see trends and anomalies in data, which can be particularly valuable in real-time environments. Visual techniques such as heat maps and tree maps, which help reveal patterns in homogenous data, were virtually unknown 10 years ago, but are used today in many places ranging from public Web sites to advanced trading applications.

Real-time environments require rapid comprehension of a dynamically changing situation - whether in the stock market, an emergency response center or an operations control center. Visualization can also help reveal patterns in complex, heterogeneous, rapidly updating data.

With a technique called visual fusion, visualization can bring together loosely related pieces of historical, real-time and planning data into a common visual framework making otherwise hard-to-detect patterns visible. An infrastructure operations visualization can depict schedules, alerts, status, performance metrics, plans and errors all within a common visual framework. Adoption of "sensors" (RFID tags, remote cameras, GPS and other monitoring devices) further increases the data volume. An operations manager has simply articulated the information overload problem: alerts and icons out of context are useless.

Even when the data is highly heterogeneous, it can still be brought together in a single visual interface based on common underlying attributes such as location, time, relationships and classifications. Interactive techniques such as selection, filtering, associations, search and aggregation can help operations staff understand and assimilate groups of events and isolate the root causes of those events. The solution can be used on a video wall, on a desktop or even on a laptop in the field. The resulting visualization provides for faster situational awareness, better management of multiple ongoing events, reduces decision latency, and fosters participation among more people in the organization. The organization has better coordination of limited resources and can stay on top of critical situations when they occur, resulting in significant cost savings.

Analysis

Visualization is extremely powerful for analysis. For example, a simple scatterplot can easily reveal patterns as a related string of dots and anomalies as clusters or outliers.

Scatterplots currently appear in many analytic visualization tools. Thanks to multitier architectures, scatterplot capacities can now display tens of millions of unique simultaneous data points, helping analysts identify opportunities, isolate risk, or find the fraudsters and the hackers. As one security administrator said, "Visualization saves me time when I review this data and helps me spot the most dangerous activities."

Powerful simulation models used in everything from risk forecasting to supply chain optimization empower businesses to create more accurate forecasts and perform what-if analysis. However, given the complexity and sheer volumes of data associated with many models, developing, understanding and communicating about the model can be difficult. With interactive visual representations of these models, not only are the results of the model more explicit, but both the modeler and the audience can interactively adjust, learn and explore the model. "Using visualization, we immediately saw that there was more basis risk than we had captured and it was shifting," said the CRO at a European bank. With visualization, the bank was able to produce higher quality data and models, with faster analysis and fewer errors helping them to reduce capital allocations by more than ten million dollars.

Visualization of non-numerical data - such as business processes, entity relationships, text, geographic data, imagery and video - represents another major opportunity. The primary challenge is to find meaningful representations that meet the users' needs. Complex tasks such as searching for a specific document, understanding relationships and sequences in a group of people or investigating a competitor's activities over time can be approached with the right visual representations. These visual representations are, in turn, dependent on the underlying algorithms (e.g., cluster analysis, automated taxonomies, extracted meta data, relevancy scoring, concept maps). The correct visual interface fits the underlying problem, as well as user skills and data models. With the correct approach, the end user benefits with a tool that can navigate through a rich, complex data set, with faster access and utilization of the data available and a higher degree of confidence in the analysis.

Broad Dissemination

Page 1 of 3.