Hacker News Reader

Comment by superdimwit | Hacker News Reader

I feel the circular plots make the data harder to read. I suspect some boring line plots with time of year on the X axis would be much more visually efficient

View on HN

I agree.

In general, circular plots are made by "designers" to have data look cool but they're not what someone actually trying to use the data would choose. I will actively caveat that for things like antenna radiation patterns, they are the natural choice.

But for data that has no relation to circles, it merely confuses the reading of the plot. It is very difficult to make a numeric judgement of a data point where both the radius and angle are changing.

Quoting Tufte probably, they lead to confusion between the areal depiction and the linear relationship of the numbers.

Pie charts, spider diagrams, fitness / health monitor circles, energy usage in your utility bill shown as circles, these plots, etc., they're mostly a graphical gimmick.

I agree. Infographics and the like are a cancer to the graphics world. I want data in its most simplest form, digest it using simple semiology and graphical grammer and move on. I find Cleveland et. al (Visualizing Data), Jacques Bertin (Semiology of Graphics) and Leland Wilkinson (Grammar of Graphics) much more 'scientific' in their approach than Tufte. Reading Tufte's books is akin to reading Josef Alber's - Interaction of Colors, which btw is the gospel of color theory amongst artists and designers: They provide interesting view of the problem but not the solution to the problem. They are coffee table books in my opinion - bedazzling and extremely interesting but useless in practice. Read Edwin Land (Kodak scientist) if you want to understand color theory.

I like the aesthetic of the circular plots and think it's very creative. Though if we were to take Mr. Tufte's advice to heart: the surface that's outlined by the colored ribbons does not increase in the same proportion as the actual number of visitors, which might influence the way the data is perceived visually. But I don't want to rain on somebody's parade who managed to make far more beautiful plots than I ever have..

I generally hate radar plots, but I thought it worked perfectly for visualizing seasons, which operate in a circle. Linear graphs of months of a year always look funny to me.

Circular graphs are half the problem I see on the "dataisugly" subreddit. At least this one isn't too bad.

Hmm, the first plot took me a long time to understand - way too much going on.

Second this, my mental calendar is a circle, but running counter-clockwise with January 1 at the 12, Feb at 11, etc. I have no problem looking at linear timelines, but viewing these clockwise representations of the year, with January at the 8 position, was surprisingly difficult.

That said, the data is really interesting.

Fascinating! I would have preferred January at the 6 o'clock, with the peak temperature & community interest (Northern Hemisphere at least) at 12 o'clock, and rotating clockwise.

It is so interesting to me that your mental calendar goes counter-clockwise. What is the intuition that picks that handedness?

I've always been curious about this -- mine is also counter-clockwise, but with January 1 down around the 7 o'clock position. I've always assumed I imported that from a calendar with aligned seasons (like with the peak of summer at July/August being at the top of the circle), but no real idea of where it originated.

Where do you put the break in the graph, then? Winter solstice, maybe?

I think for data which is circular in nature (in this case, as the Earth goes around the Sun), it makes sense.

Yes, a linear presentation works too. For me, that would be harder to read.

Reply on news.ycombinator.com