plotnine.mapping.after_stat¶
- plotnine.mapping.after_stat(x)[source]¶
Evaluate mapping after statistic has been calculated
- Parameters:
- x
str
An expression
- x
See also
after_scale()
For how to alter aesthetics after the data has been mapped by the scale.
stage
For how to map to aesthetics at more than one stage of the plot building pipeline.
Examples¶
[1]:
import pandas as pd
import numpy as np
from plotnine import (
ggplot,
aes,
after_stat,
geom_bar,
labs
)
after_stat¶
geom_bar
uses stat_count
which by default maps the y
aesthetic to the count
which is the number of observations at a position.
[2]:
df = pd.DataFrame({
'var1': [1, 2, 2, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 5]
})
(ggplot(df, aes('var1'))
+ geom_bar()
)
[2]:
<Figure Size: (640 x 480)>
Using the after_stat
function, we can instead map to the prop
which is the ratio of points in the panel at a position.
[3]:
(ggplot(df, aes('var1'))
+ geom_bar(aes(y=after_stat('prop'))) # default is after_stat('count')
)
[3]:
<Figure Size: (640 x 480)>
With after_stat
you can used the variables calculated by the stat in expressions. For example we use the count
to calculated the prop
.
[4]:
(ggplot(df, aes('var1'))
+ geom_bar(aes(y=after_stat('count / np.sum(count)')))
+ labs(y='prop')
)
[4]:
<Figure Size: (640 x 480)>
By default geom_bar
uses stat_count
to compute a frequency table with the x
aesthetic as the key column. As a result, any mapping (other than the x
aesthetic is lost) to a continuous variable is lost (if you have a classroom and you compute a frequency table of the gender, you lose any other information like height of students).
For example, below fill='var1'
has no effect, but the var1
variable has not been lost it has been turned into x
aesthetic.
[5]:
(ggplot(df, aes('var1'))
+ geom_bar(aes(fill='var1'))
)
[5]:
<Figure Size: (640 x 480)>
We use after_stat
to map fill
to the x
aesthetic after it has been computed.
[6]:
(ggplot(df, aes('var1'))
+ geom_bar(aes(fill=after_stat('x')))
+ labs(fill='var1')
)
[6]:
<Figure Size: (640 x 480)>