In two dimensions, a bivariate chart is as good as it gets for clean presentation. Bivariate means simply that two things which vary may be plotted against one another. Bivariate charts are almost always time series, which are almost meaningless.
A modification of a time series chart is a time-on-time series chart, such that the entire year wraps around on (say) a weekly basis, giving 52 wavy lines striping across a chart that is only seven days wide. This is not strictly speaking a time series chart, because what happens on one Friday is not at all connected with what happens on another Friday due to one of the Fridays being before the other. These charts illustrate the idea of Friday as a category, after Thursday and before Saturday to be sure, but the Friday-ness of results will drown out any other effect such as the January-ness, or the twenty-third-ness. This is only true, of course, if there actually is something about Fridays. A time-on-time series chart is the appropriate tool to show you a cyclical dependency embedded within time — that is, if what happens on Friday stays on Friday.
How many times have you noticed a seeming pattern in a wavy line’s waves which you immediately suspect is a weekly or payday function, but when you look at the data more closely, you cannot find any pattern to it — stock market returns frequently seem to have a recurring pattern to them, but for the most part, it ain’t weekly. Simple patterns such as this get priced in, so that pizza delivery and beer company stock prices don’t surge on Fridays and Saturdays, nor toward the end of the work day, for that matter. More sophisticated analysts (at least those who get paid to sound sophisticated) speak in terms of business cycles, but they don’t call them yearly patterns except in certain cases too obvious to be regarded as analysis — nobody needs a chart to understand that some sectors see increased sales before Christmas. The simple time series is only marginally more useful than a long list of numbers — it is essentially a one-dimensional chart, wasting space through graphic design.
Here is a nearly perfect chart of our ChinaVirus mess, with time only present as an embedded, animated function — when you look at a printout, time is almost meaningless in this chart, aside from that which can be inferred due to the fact that we are talking about growth. If this chart were labelled “populations dwindling”, we would infer that time moved the other way in this chart, without change any of the labelling.
The chart and its explanation are embedded in an excellent YouTube video from the pretty-darned-good channel MinutePhysics:
Why is this chart so useful? Because it tells you what you want to know — how we are doing compared to other data points that we have, and what we can expect to see next.
The dual logarithmic scale removes exponents from the picture, and the absence of a time axis removes coefficients and offsets from the picture. Literally, from the picture, which you may notice, is a straight line.
This is a beautiful application of fairly simple math (I am not being cute), and is one of the best examples of that famed land of desire, the “simplicity on the other side of complexity.”