Plotting averages and other aggregations on existing data¶

In [1]:

import pandas as pd
from plotnine import *

df = pd.read_csv("countries.csv")
df.head()

import pandas as pd from plotnine import * df = pd.read_csv("countries.csv") df.head()

Out[1]:

	country	continent	gdp_per_capita	life_expectancy	population
0	Afghanistan	Asia	663	54.863	22856302
1	Albania	Europe	4195	74.200	3071856
2	Algeria	Africa	5098	68.963	30533827
3	Angola	Africa	2446	45.234	13926373
4	Antigua and Barbuda	N. America	12738	73.544	77656

Plotting averages on scatter/dot plots with stat_summary¶

The easiest way to plot an average on top of an existing chart is to use stat_summary. In the example below, we use it with several parameters:

• geom=point to use a dot as the visual representation of the statistical summary
• fun_y=np.median to use numpy's median method to calculate where to place the dots. It will use the median to place the dot on the y axis (fun_y is "function for the y axis").
• color='magenta' to make it magenta, unlike the other dots

In [17]:

import numpy as np

(
    ggplot(df)
    + aes(x='continent', y='life_expectancy')
    + geom_point(color='grey', alpha=0.5)
    + stat_summary(geom='point', fun_y=np.median, color='magenta')
)

import numpy as np ( ggplot(df) + aes(x='continent', y='life_expectancy') + geom_point(color='grey', alpha=0.5) + stat_summary(geom='point', fun_y=np.median, color='magenta') )

Out[17]:

<ggplot: (312057085)>

Note that you cannot do fun_y='median', even though it seems like you should be able to.

Plotting maximum/minimums on scatter/dot plots with stat_summary¶

In the same way you use np.median to plot the median above, you can also use np.max and np.min.

In [24]:

import numpy as np

(
    ggplot(df)
    + aes(x='continent', y='life_expectancy')
    + geom_point(color='grey', alpha=0.5)
    + stat_summary(geom='point', fun_y=np.max, color='red')
    + stat_summary(geom='point', fun_y=np.min, color='blue')
)

import numpy as np ( ggplot(df) + aes(x='continent', y='life_expectancy') + geom_point(color='grey', alpha=0.5) + stat_summary(geom='point', fun_y=np.max, color='red') + stat_summary(geom='point', fun_y=np.min, color='blue') )

Out[24]:

<ggplot: (310030653)>

Anything is possible¶

Play around with it, it can get fun

In [77]:

import numpy as np

(
    ggplot(df)
    + aes(x='continent', y='life_expectancy')
    + stat_summary(geom='line',
                   fun_y=np.max,
                   color='purple',
                   group='continent')
    + stat_summary(geom='ribbon',
                   fun_ymin=np.min,
                   fun_ymax=np.median,
                   color='none',
                   fill='cyan',
                   group='continent',
                   alpha=0.2)
    + geom_point(color='grey', alpha=0.5)
    + stat_summary(geom='point',
                   fun_y=np.median,
                   color='black')
    + stat_summary(aes(label=after_stat('round(y)')),
                   geom='text',
                   fun_y=np.median,
                   color='black',
                   size=10,
                   ha='left',
                   nudge_x=0.1)
)

import numpy as np ( ggplot(df) + aes(x='continent', y='life_expectancy') + stat_summary(geom='line', fun_y=np.max, color='purple', group='continent') + stat_summary(geom='ribbon', fun_ymin=np.min, fun_ymax=np.median, color='none', fill='cyan', group='continent', alpha=0.2) + geom_point(color='grey', alpha=0.5) + stat_summary(geom='point', fun_y=np.median, color='black') + stat_summary(aes(label=after_stat('round(y)')), geom='text', fun_y=np.median, color='black', size=10, ha='left', nudge_x=0.1) )

Out[77]:

<ggplot: (312934560)>

In [ ]: