Appendix E — The ggplot2 phrasebook

ggplot2 proved that a grammar of graphics could be a working tool and not only a theory. Most readers of this book arrive from it. gog makes its own choices inside that same idea, and this appendix is the translation between them: what to reach for, and what will feel different.

It assumes the plot you want can be drawn. For whether it can, Coverage reads ggplot2’s whole reference index against the kernel and sorts the difference both ways.

E.1 Side by side

	ggplot2	gog
Languages	R	R, Python, Julia and JavaScript, one engine under all of them
Naming	`geom_point`, `aes`, `alpha`	`point`, channels by plain name, `opacity`
Where `x`/`y` live	inside each `aes()`	plot-level when written first, shared by all layers
The histogram	`geom_histogram()` or `stat_bin(geom = "bar")`	`bar * bin`
Log axis	`scale_x_log10()`	`x(gdp, scale = "log")`, written on the binding
3-D	external packages	one more position: `z`
Animation	gganimate, a separate package	`play` is a channel of the kernel: `+ play(year)`

E.2 Translations

You write in ggplot2	You write in gog
`ggplot(df, aes(x = a, y = b))`	`data(df) + x(a) + y(b) + …`
`geom_point()`	`point`
`geom_line()`	`line`
`geom_step()`	`step`
`geom_area()`	`area`
`geom_col()`	`bar`
`geom_bar()`, which counts the rows for you	`bar * count`, where the count is a word in the sentence
`geom_histogram(bins = 30)`	`bar * bin(30)`
`geom_freqpoly()`	`line * bin`
`geom_density()`	`line * density`
`geom_smooth()`	`line * smooth`
`stat_summary(fun = mean, geom = "bar")`	`bar * mean`
`aes(color = g)`	`color(g)`
`aes(size = v)`	`size(v)`
`aes(shape = g)`	`shape(g)`
`aes(alpha = v)`	`opacity(v)`
`aes(group = g)`	`group(g)`
`scale_x_log10()`	`x(v, scale = "log")`
`scale_color_viridis_c()`	`color(v) + palette("viridis")`
`facet_grid(r ~ c)`	`\\| facet(c) / facet(r)`
`facet_wrap(~ g, ncol = 4)`	`\\| facet(g, wrap = 4)`; `/` runs the same line down instead, so one number replaces `nrow` and `ncol`
`facet_*(scales = "free_y")`	`y(v, free = TRUE)`, written on the axis it frees, so there is no `free_x`/`free_y`/`free` to choose between
patchwork’s `p1 \| p2`, `p1 / p2`	`plot_a \| plot_b`, `plot_a / plot_b`: the same two operators, told apart by the operand
patchwork’s `plot_spacer()`, `plot_layout(widths =)`	`theme(width =, height =)` on the plot that wants the room; the blank corner of a marginal plot is what a shared axis leaves over
ggExtra’s `ggMarginal()`	`top / (main \| right)`, three ordinary plots (Composition)
`coord_flip()`	swap the `x` and `y` bindings: orientation is read, not switched
`labs(title = , x = , y = )`	`title()`, `x_label()`, `y_label()`, `z_label()`
`color = "red"`, `alpha = 0.5` inside a `geom_*()`	`style(color = "red", opacity = 0.5)`
`theme(…)`	`theme()`, the page rather than the ink; `style()` stays with a layer’s constants
`ggsave("plot.svg")`	the rendered plot is SVG; write it to a file from R as text

E.3 What to expect to feel different

+ is still +, and on purpose. ggplot2’s author has said he would use the pipe today, so the obvious question is why gog kept the operator he regrets. Short version: Python has no pipe to keep the four languages spelling one syntax, and a pipe makes every atom a function, which would leave bar * bin meaning nothing. Operators has the long version, including what commutes here that did not there.
Nothing is computed unless you name it. bar draws your numbers as they arrive; if you want counting, you write count. Where a geom counts for you, gog waits to be asked.
There is one spelling. No geom/stat pairs, no “specify either one.” If you can say a plot two ways here, one of them is a bug. Report it.
The scale rides the binding. You name a channel once, and its scale is a property of that binding, not a second sentence about it.
Setting is not mapping. A constant maps no column, earns no legend, and lives in style(). The quoted-string ambiguity (color("red")) does not exist here: it is refused, with the fix named.
Refusals give direction. A sentence gog cannot draw is refused, and the message names the sentence it would accept. If you are used to a warning and a plot, expect a stop instead. The errors are part of the manual: this book renders them on purpose.