I am sometimes accused, mostly by myself if I’m being honest, of seeming to think that Gaussian kernels can solve just about any data problem. This is a selection of visualizations applying kernels to one-dimensional data.
Here we have a nice simple comparison of two distributions based on data from an automated passenger counting system. One shows when people board buses, the other when they get off – the temporal lag providing a sense of how many people are riding at any given moment and for how long. More here.
Here, an image from my master’s thesis looks at relative distributions of travel speed across stop-to-stop segments of transit routes. As each segment has its own distribution over time as service repeats itself, this slices the observations on each segment into percentiles and then looks at the distribution of those percentile scores. Ghosty little lines allow you to better see how fast values are changing.
This one looks at transit stop spacing for a variety of transit agencies, calling out two in particular for comparison to the norm for the group. More here. Note that browser scrolling provides a nice elegant way of dealing with long tails.
Finally, another image related to my master’s thesis, this one comparing values at different times and different days. Note the use of negative space within the frame to annotate and explain the diagram.
Get your mind out of the gutter ;-)
There are so many more things I could include here, but I think you get the idea.