“I’ll know it when I see it.” That’s what business professionals often say when they’re digging through data, looking for an interesting trend on which to capitalize, seeking the root cause to a problem in their operations, or maybe just trying to get to know their data as they define metrics or set targets for their business.
Too often though, despite being knowledgeable about their industry and experienced in running their business, professionals may not be skilled users of visual analytics tools, and those insights can hover just out of reach. Which plots work best for which types and combinations of data? Why can’t I figure out how to make the plot I have in mind? When I do start working on a visualization, why does the system tell me “You need another dimension in the column tray” or other unintuitive messages?
Surely one needs experience, judgement, and training to be a proficient analyst, but these days it seems the flexibility and capability of tools can actually be a deterrent to getting the job done right. When the message behind a visualization isn’t clear, it’s often because the user labored over configuring details when the big picture wasn’t right in the first place. Why can’t the tool guide you to the right thing? Why would you “search” if you could “find” instead?
These are questions addressed by Spotfire Recommendations, an extremely useful new capability in the latest cloud and enterprise releases of TIBCO Spotfire. With our roots in the study of human-computer interaction and over 18 years of experience solving data problems across multiple industries, Spotfire has learned a few things about visualization best practices. Along the way, we’ve also listened to academic experts, including among others, Edward Tufte, Stephen Few, and Ben Shneiderman, the latter being our founder’s post doc advisor. Ok, whatever, you say. So how does Spotfire Recommendations actually help?
When you first open a new data set, and at all times while you are exploring, Spotfire Recommendations works in the background to profile the data and metadata in your analysis files. The distinct count of values in each column, the data types, the range of values, potentially hierarchical relationships among columns, and other characteristics are all surveyed.
When you click the Recommendations light bulb icon and select one or more columns in which you are interested, Spotfire Recommendations springs into action, using the data profile and user selections together to present a gallery of potentially useful visualizations. We can’t guarantee that a game changing insight is in the first plot but we do our best to not show visualizations that violate the rules of good data analysis.
For example, if you pick a single numerical variable, you’ll see a histogram among the views. If you pick a numerical variable and two categories, Recommendations might propose a cross tab style visualization with the numbers in the cells, and the organization of the categories in hierarchical order on the rows. Or, if one of those categories is a recognized geographic field, you might also see a geographic map with the number used for area shading and/or marker layers on top. You might also see a bar chart with the numbers as bar heights, the smallest unique valued category along the bottom and the others used to segment the bars or split them into hierarchies.
If it seems a plot is close, but not exactly what you’re looking for, you can click once to get additional similar ones, in which Recommendations has re-assigned variables to axes using the same plot construct. The point is, you will see a gallery of best practice plots, among which is likely to be what you want… and you won’t see a bunch of plots which don’t make sense for the data you’ve picked. Because the visualizations fully render your data, you will be browsing for insights instead of clicking or dragging to configure plots. This is essentially a reversal of the classical data analysis flow—we bring the visuals, you bring the insight! Just a few clicks generates a fully linked multi-faceted view of your data, as driven by your own experience and expertise, not the product user manual.
For novices, Spotfire Recommendations reduces the requirement to study the mechanical aspects of configuring visualizations (but doesn’t remove the responsibility to know your data!) and for pros, Recommendations dramatically speeds the creation of fully featured data applications.
Check out TIBCO Spotfire Recommendations and let us know how it works for you.