I was not speaking about Physics, I was speaking about experimental sciences. In doing simulations, I was happily producing tons of heatmaps. But if you are conducting experiments, getting one data point may take minutes, hours, days, or even months. This limitation often leads to the presentation of your data as X-Y dependence, which are typically the three plot types I mentioned above.
Surely it is not about penguins, I have nothing against them 
I understand surely it’s just some dataset, that could be about stars or minerals just as well. The issue is: it is a multidimensional dataset. The data complexity needs attention, thus saturating part of the bandwidth. On the other side, this data complexity is not initially needed for the explanation of the basic concepts.
No, I am definitely not going to start a blog on AoG. For many reasons. Also, note, Bogumił Kamiński is the author of DataFrames.jl