Synapse

Using Design Patterns to Find Greater Meaning in Data

This blog originally was originally published on O’Reilly.  

At a time when businesses across every industry are debating how to transform their data into actionable assets, analyzing design patterns through visualizations offers a clearer picture of your data and helps to handle the increasing complexity of your business today.

Design patterns are one of the best ways to solve commonly recurring problems. In design and engineering, a design pattern provides guidance for how to solve a problem with a repeatable solution—a template approach that can apply to different types of situations. Your data sets may evolve and your questions may become more specific, but you still need an approach to navigate and analyze your data in a more structured way.

Patterns based on function can enable you to identify differences and similarities more clearly, understand relationships and behaviors more intimately and predict future results with a greater level of certainty. When these patterns are presented as visualizations, they help you 1) see comparisons, 2) make connections and 3) draw conclusions from your data sets.

The three major functions to data visualizations are based on our own analysis of charts for their functional purpose, which ultimately can be described as the following:

  • Comparisons: Visuals based on comparisons can help reveal differences and similarities (e.g., visuals that show rank order or how events unfold over a specified period of time).
  • Connections: Linked and related data produce visualizations that reveal themes or associations within the data set (e.g., categorizing data into logical groups).
  • Conclusions: Visualizations can  provide answers to questions or, more directly, present an outcome. For example, what-if scenarios can illustrate the results of a decision.

To ground these design patterns, let’s review a scenario that links them into a plausible series of questions using one hypothetical data set and sample data visualizations from each major function.

Consider the process of evaluating a range of investment options. First, you may review the investment performance. Deep comparisons show not just the returns, but also the sequential path that has created those returns. You may ask: Which fund(s) has provided the best returns? And how does that compare to the benchmark?

SHOWING COMPARISONS: Fund & Benchmark Performance

Figure1_v2

Figure 1: Bar with sparkline.

As shown in Figure 1, the bar chart with sparkline enables you to review the data at two different levels: a high-level assessment of the short-term three-month returns is represented with the bar chart, while the sparkline (the line chart below the bar) provides the details of the historical returns. Quickly and concisely, the sparkline shows the path that has led up to the most recent returns. You can then assess that a narrow path provides consistent returns across the years while a wide path provides varied returns. Side-by-side comparisons of funds organized into two columns , % Returns and % Ahead of Benchmark, enables peer comparisons and fund-specific benchmark comparisons. Hence you can see that not only has Global Large Cap Core provided positive returns, it has also provided the best and most consistent returns when compared to the benchmark.

Next, you may ask about how the returns relate to overall category performance. This is a time to step back and gain a higher level of context.

MAKING CONNECTIONS: Category Performance

Figure2_v2

Figure 2: Symmetrical area profiles.

The string of charts in Figure 2 are 10-year to year to date (YTD) performance returns, which can be interpreted as individual charts or a group of category charts.
Similar to sounds waves, the symmetrical area charts grow equidistant from the source (the zero line) at each time interval to accentuate the returns even further. Here, the y-axis is shown in percentage. Instead of using the zero line to indicate positive or negative returns, it uses color to denote if the category returns are positive (black) or negative (red). For example, Multi Cap Russell 3000 Growth produced 20% positive returns within the 1-year time period and is shown with color fill in both directions from the zero line to purposefully duplicate the large gains and specifically uses black color fill to indicate the returns are positive. As evident from the name, the symmetrical chart doubles the returns to emphasize the amount with color fill.

What else can you derive from organizing the information in a spectrum of negative to positive returns?

Based on this organization, three groups of categories have resulted in straight losses (red), heavy gains (black) or a mix of gains and losses across a decade of returns. The string of charts makes it easier for you to see these three groups of categories to assess their distribution. Just like sound waves, each chart is a sound bite that streams the returns for each category with a scream announcing a huge gain (e.g., Multi-Cap Russel 3000 Growth) or loss (e.g., Mid Cap Russel Mid Cap Growth). In some cases (e.g., Large Cap S&P 500), the chart quietly announces mixed returns to adequately demand less attention.

Next, you may wonder how you would have fared if you had invested in certain funds. You may ask: If I had purchased this fund five years ago, what would my return be? And what about the YTD returns? Since market timing is key to investment choices, the following presentation of hypothetical investments represents a range of results.

DRAWING CONCLUSIONS: Fund Performance

figure3

Figure 3: Curved line paths.

In Figure 3, varied performance results become clear with a layered approach to show five potential entry points (10-year, 5-year, 3-year, 1-year, YTD) into an investment. For example, the International Large Cap Core fund provided 27% YTD returns, which contrast the negative returns you would have received had you invested in the fund 1, 5 or 10 years ago. Here, conclusions are derived based on known inputs with a divided review of positive or negative outcomes (shown on the y-axis).

The line weights help to identify each entry point and show the range of differences between the entry points. Accordingly, resulting returns are shown with simplified curves that connect the inputs and outputs. In this case, the chart has been customized to show an instance in which the user has opted to see the YTD return values as percentages listed to the right of each resulting output.

Combining and altering design patterns to yield simple, effective visuals

The sample visualizations shown above combine common design patterns, like the bar and sparkline (as seen in Figure 1), and alter the traditional line and area charts to simplify the visual and effectively highlight the main points. Doubling the area chart (as seen in Figure 2) emphasizes the returns and clearly distinguishes the groups of category charts. Curved lines (as seen in Figure 3) link two points  (the entry point and the performance result) to present final outcomes.

Although the line and area charts are commonly used and familiar to most people, these alterations require some additional effort to fully decode the meaning. Once explained, you can apply the pattern of use to more of the same charts. And while the Symmetrical Area Profiles and Curved Line Paths are new visualization patterns, increased adoption will make these two new patterns more familiar, common and quickly interpreted.

Selecting visualizations based on functional purpose

Think of visualizations as functional design patterns that can be reapplied to your own data sets. A design pattern provides a template approach that can apply to different scenarios across your business including portfolio construction, performance attribution, risk management, and many others. The key is to identify the best-matched template and then apply it based on the functional aspects of the business needs. To start on this path, navigate your data with visual inquiries that help you see comparisons, make connections and draw conclusions.

In most cases, a deeper understanding of your data will require further investigation with multiple visualizations that can aggregate, filter and drill down into granular details. But at the very least, the visualizations provide starting points that can lead you to ask more pointed questions, explore the data further and ultimately follow a path to gain more valuable insights.

For More Information, Check Out Our New Book

Visualizing Financial Data

“Anyone who has tried to design a display of financial data will learn enormously by studying this book.”
– Amazon review by Paul Kahn, Experience Design Director

Authors

Julie Rodriguez – Collaborative Creative, Visionary Thinker, Creator of vizipedia.com
Julie is a collaborative creative with a passion for helping clients visualize, define and achieve their business goals. She provides user research, strategy and UI design to create best-in-class digital experiences that satisfy business objectives and respond to user needs. Julie has patented her work in commodities trading and data visualizations for MATLAB and is the creator of vizipedia.com, a data visualization pattern library used to increase visual literacy and knowledge share for the data visualization community.

Piotr Kaczmarek – Associate Creative Director
Piotr Kaczmarek is an information designer with over 20 years of experience in qualitative and quantitative data visualization, presented in formats ranging from print and digital displays to interactive animations. He is the co-author of Visualizing Financial Data, a book about visualization techniques and design principles that includes over 250 visuals depicting quantitative data.