Comment by minimaxir | Hacker News Reader

Comment by minimaxir | Hacker News Reader

ggplot2 is the primary reason to use R. Not that Python's modern charting libraries like bokeh/seaborn are bad, but ggplot2 is so robust and customizable while requiring very little LOC. And that doesn't even include external libraries which extended ggplot2, including pairs plots (https://rdrr.io/cran/GGally/man/ggpairs.html) and automatic interactive charts with plotly (https://plot.ly/ggplot2/)

ggplot 2.2.0 added a lot of customization features which make it easier to make a plot look more unique than the stereotypical ggplot2 chart, which I plan to cover in a tutorial soon. (here's an older ggplot2 tutorial of mine which still holds up: http://minimaxir.com/2015/02/ggplot-tutorial/ )

I think the biggest issue with matplotlib is it's straddling of the fence between MATLAB and pythonic syntax throughout the library. It makes it actually pretty hard to predict what the syntax/call signature of a particular method is if you don't use it fairly regularly.

This. Matplotlib is easily the most jarring library I use on anything like a regular basis. I really want a requests-for-charting, which has a more pythonic syntax.

seaborn is actually pretty nice, though not as diverse.

There's also altair[1], which also has a grammar of graphics. Most visualizations need one line of code, and it integrates nicely with pandas:

    Chart(df).mark_point().encode(
    x='year', y='horsepower', color='brand')

[1]:https://altair-viz.github.io

plotnine is an excellent port of ggplot2 to python.

https://plotnine.readthedocs.io/en/latest/index.html

Reply on news.ycombinator.com