Terrorist attacks. Insurance fraud. Product recalls. The headlines are rife with examples where the ability to explore data, “connect the dots across data sources,” obtain relevant, timely analysis and share the results could prevent serious harm from happening to people and organizations.
Data analysis plays an important role in virtually every industry. Law enforcement agencies, for example, need to identify criminal activity, businesses need to analyze financial performance, marketers need to understand buyer behaviors and healthcare organizations need to identify insurance fraud – to name just a few.
The Analysis Challenge
The fundamental goal of analysis is to turn data into knowledge that can be acted on by decision makers. Accomplishing this task, however, poses a daunting challenge for most organizations as the volume of data and the velocity at which it pours into organizations is unprecedented. The information overload is keenly felt by analysts.
Website metrics, public domain sources, RFID devices, point of sale data, customer loyalty programs, performance management applications, traditional operational databases and more yield ever growing mountains of data. As the data repositories continue to grow at a record pace, the window of time to assimilate, comprehend and act on the resulting knowledge continues to shrink. Deluged with data, analysts face an almost insurmountable task.
Unfortunately, traditional business intelligence (BI) solutions are not designed for this modern analysis environment, and are simply not getting the job done. Rigid in nature, they can be hard to install and difficult to learn. Most were built for online reporting, deliver basic dashboards of operational performance and rely on underlying data models. They lack the highly interactive visualization capabilities needed to rapidly uncover hidden relationships in data. Further, they fail to provide the ability to integrate data and share results.
Clearly, there is increasing demand for tools that allow analysts to find relevance in a growing sea of data. Over the past few years, most of the innovation in analytics has been in the area of automated information analysis, a technique that attempts to reveal relevant insights automatically. We are, however, nowhere near the point where we can remove the analyst from the equation. In fact, the most important component in the analytic process, when searching for the unknown, is human judgment. It is still the analyst who makes the discovery. Successful next-generation products must therefore deliver highly effective innovation that is analyst-centric.
Next-generation BI products must address the shortcomings of the past generation and allow analysts to rapidly assimilate, comprehend and act on all of the information at their disposal - even when they don’t know what questions to ask.
Interactive Analytics (a new form of data visualization) delivers on this promise. The human brain has unparalleled pattern-recognition capabilities. Visualization uses visual metaphors to support our cognitive operations, enabling us to explore and understand data by enhancing our ability to detect patterns, relationships and anomalies. As an integral part of the analysis, visualization supports the ability to pose questions through direct interaction with the displays that are rendered.
Interactive Analytics holds great promise for quickly and effectively detecting relevant insights, and is based on highly interactive visualizations that allow analysts to rapidly identify and comprehend non-obvious patterns within the data. These visualizations include relationship graphs for link analysis, charts and heat maps for quantitative analysis, timelines for temporal analysis, and maps for geospatial context. Supported by powerful functions to integrate and manipulate data and share results, this approach yields actionable intelligence within compressed time frames.
Three emerging innovations that support Interactive Analytics are Interactive Visualization, Unified Data Views and Collaborative Analysis. These technologies comprise the pillars of this new approach, a human-centric approach to data analysis.
Most analysis tools require one to know what one is looking for in advance, such as: “What is my revenue by region?” or “How many customers do I have?” This is contrary to navigating an information space and determining what is of interest on the fly. Today, businesses need to discover the unknown.
Relationship graphs, which fall under the science of link analysis, are is used to discover and understand relationships between seemingly unrelated entities. Link analysis can be used to visualize virtually any data set and is widely being used in social network analysis as well as other applications and industries, such as showing linkages between people-products-stores or patients-drugs-symptoms. The relationship graph below, for example, shows linkages between products (blue shield) and customer transactions (green money). The analyst has isolated specific products for which there are ten or more linkages to transactions, products, or both. This type of visualization allows analysts to draw definitive conclusions that individual charts would not reveal, such as:
- The volume of transactions varies by product
- Buying behavior in the form of single and multi-product transactions differs
- Individual transactions include one, two, three or more products
Figure A. Linkages Between Products and Customer Transactions
Armed with this intelligence, the analyst may want to grow this picture to include customers, promotions, lifestyle cluster codes and other characteristics useful in understanding business performance. She may want to explore regional variations in buying behavior, profit margins, store performance and customer value. Since Interactive Analytics empowers the analyst to explore data in an unconstrained way, these questions can easily be answered.