In [2]:

import pandas as pd
from plotnine import *

df = pd.DataFrame([
    { 'lang': 'jp', 'word': 'たこ焼き', 'number': 300 },
    { 'lang': 'cn', 'word': '火鍋', 'number': 400 },
    { 'lang': 'ko', 'word': '김치', 'number': 200 },
])

import pandas as pd from plotnine import * df = pd.DataFrame([ { 'lang': 'jp', 'word': 'たこ焼き', 'number': 300 }, { 'lang': 'cn', 'word': '火鍋', 'number': 400 }, { 'lang': 'ko', 'word': '김치', 'number': 200 }, ])

Displaying East Asian text (Chinese, Japanese, Korean) in plotnine¶

When you try to display Chinese, Japanese, or Korean text on a plotnine graphic, you might get an error like this:

/Users/username/.pyenv/versions/3.9.7/lib/python3.9/site-packages/matplotlib/backends/backend_agg.py:240: RuntimeWarning: Glyph 12383 missing from current font.
/Users/username/.pyenv/versions/3.9.7/lib/python3.9/site-packages/matplotlib/backends/backend_agg.py:240: RuntimeWarning: Glyph 12371 missing from current font.
/Users/username/.pyenv/versions/3.9.7/lib/python3.9/site-packages/matplotlib/backends/backend_agg.py:240: RuntimeWarning: Glyph 28988 missing from current font.

This is because many fonts do not support these characters. To display them, you'll need to find a font that works for them.

Using Arial Unicode MS¶

If we have a dataset of all three languages, Arial Unicode MS is a good option that supports characters around the world. We'll use theme(text=element_text(family='Arial Unicode MS')) to set all of the text in the graphic to be this font.

In [14]:

df

df

Out[14]:

	lang	word	number
0	jp	たこ焼き	300
1	cn	火鍋	400
2	ko	김치	200

In [15]:

(
    ggplot(df)
    + aes(x='word', y='number')
    + geom_bar(stat='identity')
    + theme(text=element_text(family='Arial Unicode MS'))
)

( ggplot(df) + aes(x='word', y='number') + geom_bar(stat='identity') + theme(text=element_text(family='Arial Unicode MS')) )

Out[15]:

<ggplot: (310861300)>

Using ONLY Chinese or Japanese character sets (SimHei, Heiti)¶

Usually you only have one language in your dataset. Instead of using a font that has every single language in it, you might pick something more specific. For Chinese, there are specific fonts on both OS X and Windows that are built-in and work well.

• Windows: SimHei
• OS X: Heiti TC

They won't display Korean correctly, but they'll be fine for Chinese and Japanese.

In [18]:

# I'm on a mac, so I sue Heiti TC

(
    ggplot(df)
    + aes(x='word', y='number')
    + geom_bar(stat='identity')
    + theme(text=element_text(family='Heiti TC'))
)

# I'm on a mac, so I sue Heiti TC ( ggplot(df) + aes(x='word', y='number') + geom_bar(stat='identity') + theme(text=element_text(family='Heiti TC')) )

/Users/soma/.pyenv/versions/3.9.7/lib/python3.9/site-packages/matplotlib/backends/backend_agg.py:240: RuntimeWarning: Glyph 44608 missing from current font.
/Users/soma/.pyenv/versions/3.9.7/lib/python3.9/site-packages/matplotlib/backends/backend_agg.py:240: RuntimeWarning: Glyph 52824 missing from current font.
/Users/soma/.pyenv/versions/3.9.7/lib/python3.9/site-packages/matplotlib/backends/backend_agg.py:203: RuntimeWarning: Glyph 44608 missing from current font.
/Users/soma/.pyenv/versions/3.9.7/lib/python3.9/site-packages/matplotlib/backends/backend_agg.py:203: RuntimeWarning: Glyph 52824 missing from current font.

Out[18]:

<ggplot: (309137818)>

Other fonts and character sets¶

To figure out all of your other options, it's usually best to display a list of supported fonts.

In [ ]: