Comprehensive Guide to Plotnine: Installation, Data, Aesthetics, Geoms, Stats, Scales, Positions, Coordinates, Facets, and Themes

plotnine is the Python implementation of the Grammar of Graphics, offering functionality almost identical to ggplot2 for creating layered graphics.

Installation

Using pip (with Tsinghua mirror for speed):

#指定清华源快速安装plotnine
pip install plotnine -i https://pypi.tuna.tsinghua.edu.cn/simple

Using conda:

conda install plotnine

Using git on Linux:

git clone https://github.com/has2k1/plotnine.git
cd plotnine
pip install -e .

Data (datasets)

plotnine provides several built‑in datasets accessible via plotnine.data . Example to list them:

#内置数据集
print(dir(plotnine.data))

'diamonds', 'economics', 'economics_long', 'faithful', 'faithfuld', 'huron', 'luv_colours', 'meat', 'midwest', 'mpg', 'msleep', 'mtcars', 'pageviews', 'presidential', 'seals', 'txhousing'

Aesthetic Attributes

Map data columns to visual properties using aes . Example mapping displacement and highway mileage with colour by class:

#aes中设置点的属性，按class使用不同颜色
ggplot(mpg, aes('displ', 'hwy', colour = 'class')) + geom_point()

Geometric Objects (Geoms)

All geoms start with geom_ . Example listing available geoms and creating a boxplot:

print(len([i for i in dir(plotnine.geoms) if i.startswith('geom_')]))
print([i for i in dir(plotnine.geoms) if i.startswith('geom_')])
#geom_boxplot()绘制箱图
ggplot(mpg, aes('class', 'hwy')) + geom_boxplot()

Statistical Transformations (Stats)

Stats start with stat_ . Example listing them and summarising mean values:

print([i for i in dir(plotnine.stats) if i.startswith('stat_')])
#统计每组数据均值，红点表示
ggplot(mpg, aes('trans', 'cty')) + geom_boxplot() +
    stat_summary(mapping=None, data=None, geom='point',
                 fun_data='mean_cl_boot', colour='red', size=4) +
    theme(axis_text_x = element_text(angle=90, hjust=1))

Scales

Scales map data values to visual channels. Example using shape and size aesthetics with custom scales:

#点按f1使用不同marker，按cty使用不同siz
ggplot(mpg, aes('displ', 'hwy', colour='class')) +
    geom_point(aes(shape='fl', size='cty')) +
    scale_shape() + scale_size()

Position Adjustments

Control how geometries are positioned. Examples of stacked, filled, and dodged bar plots:

print(ggplot(mpg, aes('cty', fill='class')) + geom_bar())
print(ggplot(mpg, aes('cty', fill='class')) + geom_bar(position = "fill"))
print(ggplot(mpg, aes('cty', fill='class')) + geom_bar(position = "dodge"))

Coordinate Systems

Available coordinate systems include coord_cartesian , coord_flip , etc. Example listing them:

print([i for i in dir(plotnine.coords) if i.startswith('coord_')])

Faceting

Facets split data into multiple panels. Example using facet_wrap and facet_grid :

#facet_wrap
(ggplot(mpg, aes(x='displ', y='hwy')) + geom_point() + facet_wrap('class') + labs(x='displacement', y='horsepower'))
#facet_grid
(ggplot(mpg, aes(x='displ', y='hwy')) + geom_point() + facet_grid('drv ~ .', labeller = 'label_both') + labs(x='displacement', y='horsepower'))

Themes and Sub‑Plotting

Various themes are available (e.g., theme_xkcd , theme_dark ). Example creating four sub‑plots with different themes and coordinates:

from matplotlib import gridspec
p1 = ggplot(mpg, aes('displ', 'hwy')) + geom_point() + geom_smooth() + theme_xkcd()
p2 = ggplot(mpg, aes('displ', 'hwy')) + geom_point() + geom_smooth() + coord_cartesian(xlim=(5,7)) + theme_dark()
p3 = ggplot(mpg, aes('cty', 'displ')) + geom_point() + geom_smooth() + theme_matplotlib()
p4 = ggplot(mpg, aes('displ', 'cty')) + geom_point() + geom_smooth() + coord_fixed() + theme_linedraw()
fig = (ggplot()+geom_blank(data=mpg)+theme_void()).draw()
gs = gridspec.GridSpec(2,2)
ax1 = fig.add_subplot(gs[0,0])
ax2 = fig.add_subplot(gs[0,1])
ax3 = fig.add_subplot(gs[1,0])
ax4 = fig.add_subplot(gs[1,1])
_ = p1._draw_using_figure(fig,[ax1])
_ = p2._draw_using_figure(fig,[ax2])
_ = p3._draw_using_figure(fig,[ax3])
_ = p4._draw_using_figure(fig,[ax4])
plt.tight_layout()
plt.show()

Reference: plotnine documentation