In Spotfire Graphical Table visualization, the use of sparklines is a fantastic way to quickly visualize our data in table format. But, what if we have highly variable data in which it would be better to use a logarithmic scale on the Y-axes? Note, there is no option for using a logarithmic scale on the spark line axes visualizations. We have two options here: to use multiple scales or write a custom expression with multiple scales.
One of the main uses of sparklines is to show the “shape” of our data. If our data range is less variable, then a single arithmetic scale for all sparkline axes is fine. However, in the case below we need to use a different arithmetic scale for each spark line in the column to honor the high variability of the data.
Go to Properties of Graphical Table > Axes and select spark line column as seen below. Now select Settings button for that spark line column.
Then, select Axes and change radio button under “Y-axis scale” from “One scale for all sparklines in this column” to “Multiple scales.” We do the same for all spark line columns with highly variable data we want to “compare,” as seen in the second spark line column in the Graphical Table below.
Note the huge improvement in being able to see the “shape” of our data with “Multiple scales” selected, compared to the first visualization above.
Next, if we want a Log or Logarithmic scale, we can easily write a custom expression as seen below by right mouse clicking of Y-axis name in Sparkline Settings.
Insert Log function Custom Expression, then hit Okay.
Compare final Log Scale Graphical Table below to previous two arithmetic Tables above. Note visualization below has “Multiple scales.”
Finally, if we use “One scale for all sparklines in this column” instead of “Multiple scales” the results may not show enough differentiation especially if you have extreme outliers. Compare this last log image with our first arithmetic one.
Paul is a geoscientist specializing in data analytics and visualizing a lot of data in a small amount of space. As Edward Tufte says, “There is no such thing as data overload, just a failure of design.”