Smart Analytics and Data Visualization

Data visualization is a powerful way to tell a story and empower data-driven conversations. However, too often people are presented with clutter of visual charts and dashboards that do not directly resonate with them. eSimplicity puts a strong focus on human-centered design to understand our audience. We then add smart analytics capabilities that are emerging from several open sources as well as leading self-service analytics and visualization products such as Tableau and Qlik to automate insights and visualizations to the users.

 

The SIMPLE Experience

Applying human-centered design to data analytics & visualization

We build user personas and ask ourselves: Are we developing our dashboard for a salesperson with 15 seconds to spare for key performance indicators, or for a team to review quarterly data over several hours? With thoughtful planning we evaluate proper display size, and appropriately plan for fast load times. We review charts from real data and gain insights for the chart design that draws from the “sweet spot” of visual cues, for example. We are selective of color quantity, incorporate interactivity to encourage exploration, and consider progressive formatting. We continue to refine the dashboard through usability testing, put the onus on tooltips, help showcase the story within the user’s story.

Smart analytics capabilities

The trend toward self-service analytics has been a hugely positive development in the evolution of data-driven decision-making, but it has carried us only so far. To take data analysis to an even broader audience and to speed and deepen analyses for existing users, eSimplicity uses emerging smart capabilities in the areas of data prep, data discovery and analysis, NL query and automated predictive capabilities that are recently added to data analytics and visualization platforms such as Tableau, Qlik or Power BI. These solutions are applying automation and behind-the-scenes decisions to inherently complex processes, such as the selection of algorithms. Note that the danger is that “black box” (nontransparent) predictive systems may not make decisions or recommendations that are in the best interests of the organization. Hence, it requires experience and awareness to switch some of these smart capabilities to manual control. Our team review these smart predictions to ensure that they do not lead to potentially costly actions. One approach is to let analysts and data-savvy business users experiment with automated prediction features and then share their findings with data scientists, who can enhance the analyses and put carefully curated models into production.

Explainable but not magical predictions

Trust and transparency are two of the biggest issues that organizations will face as they embrace ML and AI technologies. Expect change management and training to be required in order to promote trust in ML- and AI-based recommendations and suggested actions. People will more readily accept computer assistance and accept changes in processes if they understand how and why decisions and recommendations are made. That’s where transparency comes in. Smart systems should be explainable, not magical.