Sears Merritt, VP & Chief Data Scientist
Sunday 20 January 2019
Published with Permission from CIOReview
As data and algorithms continue to make their way into the deepest, most complex corners of business decision making, the magnitude of their impact skyrockets. Users and stewards of these powerful tools have an obligation to understand their inner workings at the most fundamental levels. While there are many ways to develop such an understanding, visual tools depicting complex relationships and conditional dependencies of data and algorithms have proven to be the most effective. In this article, we describe data visualization and two common use cases found in financial services. We also comment on three of the determinants that have supported the proliferation and evolution of data visualization methods, frameworks, and products. We close with comments about the future of data visualization and current trends.
Data visualization is an interdisciplinary field, leveraging art, science, and mathematics, and focusing on the creation and study of the visual representation and display of data. The field's rich history has evolved with technology over hundreds of years. A classical example is Charles Minard's late 1800's diagram of Napoleonic France's invasion of Russia. Methods for visualizing data can range from simple tables and two dimensional bar charts, to rich, interactive, multi-dimensional systems. Effective use of visualization techniques communicates information and highlights complex patterns efficiently. When used in a decision-making setting like healthcare, insurance, or scientific research, effective visualization enables outcomes of higher quality.
Leading data-driven firms across all industries now rely on data visualization as a means for attracting and retaining customers as well as enabling or enhancing critical decisions. In financial services, this includes helping customers plan for major life events like retirement and starting a family. Due to their long time horizons and levels of uncertainty, such events can be complex and require long-term financial planning that is not easy to reason about mentally or with traditional tables and figures. Interactive calculators and dashboards allowing one to model various scenarios and test assumptions are proving to be most useful to the modern consumer. Visualizations contained in these types of tools dynamically display how cash flows and assets change in value through time using stacked area charts or stream graphs.
Another critical function within financial services that relies heavily on data visualization is fraud detection. Fraud is typically performed by groups of individuals that have indirect and often times hidden social ties. These ties form the basis for covert communications and strategic planning and execution of criminal activities, particularly against insurance policyholders. To combat these activities, fraud analysts turn to social networks to mine for these hidden links. Visualization techniques that display network communities are used to enable efficient searching and detection of such links.
The current transformation in data visualization has been enabled by at least three arguably obvious and common factors: availability of complex and massive data sets; maturing standards; and availability of computing power, particularly on local devices. It is worth noting that these factors are also associated with the broader digital transformation currently taking many industries, including financial services, by storm.
The need to efficiently understand and extract information contained in large, complex data has grown with its availability. As a result, researchers and product innovators have invested time and capital in exploring new methods for visualizing information. This has led to an explosion of new products and platforms in the visualization space as well as substantial changes to traditional, legacy offerings.
Of course, processing and visualizing complex data requires sufficient amounts of computing power. CPU's and GPU's have become and are expected to continue to be affordable and powerful sources of computation available on nearly all electronic devices. This trend combined with method innovations that are implemented via cross-platform standards, has allowed the aforementioned products and frameworks to operate ubiquitously.
As data and analytics strategies continue to evolve and adapt to the demands of consumers and businesses alike, we expect an increasing need to continually enhance and improve human decision-making through the use of visual tools. As a result, methods, standards, and computing power are likely to continue to play a dominant role in future changes in data visualization. Hints supporting this hypothesis are already evident in emerging technologies and changes to social norms of professions like journalism. This profession is now experiencing an emerging trend that more evenly blends data analysis and visualization of trends with investigative reporting and storytelling. Mixed-reality systems leverage interactive visualization of data to act as sophisticated aids for decision-making in physical environments as diverse as oil rigs and operating rooms. Given its rich history, ability to reach broad audiences, and critical role in decision making and information dissemination, data visualization will continue to play a key role in the data and analytics strategies of leading firms.