It's a different beat, but R + dplyr + ggplot2 (or the built-in lattice and plot libs) are so much expressive, concise and efficient than other things like Python's matplotlib. The challenge is to get started with R. But then, every time you need to use another language (because it needs to run in another context, another scale) I regret the ease and fluent code dplyr+ggplot2 provide to explore datasets visually!
You can get an idea here (or google it, there are so many resources!) ramnathv.github.io/pycon2014-r/visualize/ggplot2.…
The fact is that it's based on the ifluencial grammar of graphics principles.
And other libraries use the same principles. The idea is to separate your data from the geometries and aesthetics features. Vega is a set of specs (which I never had to use but looks really great, similar to ggplot2) and has ports is various languages, one of them being javascript: github.com/vega/vega-lite