python Stacked area chart Bokeh

Question

I am trying to create a stacked area chart, which shows the number of customers by country.

So my data frame is:

date        people  country
2021-11-18  509     USA
2021-11-18  289     France 
2021-11-18  234     Germany
2021-11-18  148     Poland
2021-11-18  101     China

I don't understand how to edit the graphic design (color).

table.groupby(['date','country'])['people'].sum().unstack().plot(
    kind='area',
    figsize=(10,4))

Also I tried to use the Bokeh library for nice visualization, but i don't know how to write the code

Thanks for your help. It's my first post. Sorry if I missed something.

Answer 1

I think your are looking for varea_stack()-function in bokeh.

My solution is based on the varea_stack-example which is part of the official documentation.

Let's assume this is your data (I added on day):

text = """date        people  country
2021-11-18  509     USA
2021-11-18  289     France 
2021-11-18  234     Germany
2021-11-18  148     Poland
2021-11-18  101     China
2021-11-19  409     USA
2021-11-19  389     France 
2021-11-19  134     Germany
2021-11-19  158     Poland
2021-11-19  191     China"""

First I bring the data in the same form of the example:

from io import StringIO
import pandas as pd

df = pd.read_csv(StringIO(text), sep='\s+', parse_dates=True, index_col=0)

df = df.groupby(['date','country']).sum().unstack()
df.columns = df.columns.droplevel(0)
df.index.name=None
df.columns.name=None

Now the DataFrame looks like this:

            China  France  Germany  Poland  USA
2021-11-18    101     289      234     148  509
2021-11-19    191     389      134     158  409

Now the rest is straight forward. If your index is a DatetimeIndex you have to modify the x_axis_type of the bokeh figure. Id did this for the plot below.

from bokeh.palettes import brewer
from bokeh.plotting import figure, show, output_notebook
output_notebook()

n = df.shape[1]
p = figure(x_axis_type='datetime')
p.varea_stack(stackers=df.columns, x='index', source=df, color=brewer['Spectral'][n],)
show(p)

The output lookslike this:

You can redefine the color using the color-keyword if you like.