Why use ggplot2

Note: The ggplot2 wiki is no longer maintained, please use the ggplot2 website instead!

Reasons to use ggplot2 contributed by ggplot2 users.

• Automatic legends, colors, etc.
• The “default” output is much nicer than with base graphics, unless starting from the output of a statistical analysis, where people have written nice things in base.
• Easy two-colour gradients to distinguish positive and negative values in GAM surfaces.
• Easy access to ribbons with transparency, for confidence intervals.
• Combine multiple data sets into a single graph with a snap-together, building-block approach.
• Large variety of customizable smoothing overlays, including loess, [list the other R packages whose smoothing you support, and emphasize the 2D (kde?)].
• Handsome default settings.
• Approach your graph from a visual perspective rather than a programming perspective.
• Turn a Cartesian graph into a polar graph with a single statement.
• Store any ggplot2 object for modification or future recall.
• easy superposition (boxplot + points + lines + …) + + +
• easy facetting ++
• easy legend ++
• object oriented graph (I’m not sure for this…)
• beautiful default parameters
• theme support
• Facile facetting
• Luscious layering
• easy to make pretty and elaborate graphs
• carefully chosen default colour schemes and layout so the user needn’t worry about aesthetic values that he wouldn’t necessarily know himself well enough
• consistent look across graphics
• literate programming: you describe the plot almost in a natural language. It makes it much easier to reuse code and come back to a graph in a few years time without feeling lost.
• only few high-end functions, but with very wide applicability
• based on grid, the preferred and recommended way to develop new graphics in R
• powerful layer system that makes it easy to combine different sources of data
• superpose several layers (possibly from different data sources) on a plot with automatically adjusted common scales. In base graphics, if you want to plot something as simple as points on a map, you have to plot the map, know the x and y limits and plot the points with the exact same limits, using several trick to superpose the plots, suppress the plotting of axes in the second plot etc. With ggplot you just need to add two geom calls: incredibly straightforward and intuitive. That said, ggplot behaviour (superpose plots and adjust scales) is the default behaviour of all other software packages I know (MATLAB, Scilab, etc.) so it may well be that base R graphics are to blame here, rather than ggplot being to praise. If you want to attract R users, then this is an advantage, otherwise, this is just common practice.
• flexibility, intuitiveness, and logic of the mapping between the data and its representation. Once one gets his mind wrapped around the grammar of graphics concepts (and particularly the aesthetic mapping), figuring out how to best represent a dataset is much easier than with other graphical representation methods. For me at least, there is a kind of graphical construction of the plot going on in my head before the final code/representation. I imagine drawing lines with a pen between data columns and graphical parameters, like in a children’s game: “so data$foo goes with x, data$bar goes with y. OK. Then data$value could go with colour… oh but what about linking it to size?” etc. This process, I think, helps me constructing plots that are perceptively better or complex but still intuitive.
• In addition, ggplot holds the hand of the user by proposing great graphical representations (geoms) and easy transformations.
• visual aspect. ggplots look freakin’ cool! Even though I think it got a bit worse in the last versions (title and axes titles font size and positions not as harmonious as before, strange color choice for the default gradient scale — from pale blue to pale beige — which does not stand against the background, and a few other small things) ggplots still look very nice. The light grid on grey background helps reading the charts and makes plots more “consistent” in the middle of text paragraphs. The axes scales in darker grey and titles in black look very good and tasteful. The legends are much better than anything else I know about. Etc.
• I think the big strength of ggplot2 is graphic-artist quality default output. An an information-focussed point of view.
• Makes doing the right thing easy, while keeping harder things possible.
• ggplot2 is an intuitive library that is founded on a logical mapping between data and graphical elements, building plots up not by thinking in terms of dots on paper, but in terms of what the data means. Think about your data and what you want to communicate, not the mechanics of drawing.