CODEWITHSUNDEEP
pythonbokeh
theQman
theQman

Reputation: 1780

Using `line` vs `multi_line` for plotting many (differently colored) lines in bokeh

I have a pandas dataframe `df' in tidy format, like so

date          population   country
Feb. 1 2000   99999        Canada
Feb. 1 2000   98765        Spain
Feb. 2 2000   99998        Canada
...

I would like to do a line plot using Bokeh, where each country gets its own line and color.

One way to do this seems to be to use the legend keyword to line() to give me a different line for each country:

source = ColumnDataSource(df)
plot = figure(...)
plot.line(x='date', y='count', source=source, legend='country')

Unfortunately, it seems like there's no straighforward to select colors for each country...

And since there's a multi_line() plotting function, this seems like what I should be using. However, I don't know a simple way to do this. Something like the following can work:

plot.multi_line(xs=[df['date'], df['date']], 
                ys=[df[df['country']=='Canada'], 
                    df[df['country']=='Canada']], 
                colors=['red', 'blue'])

This doesn't seem very elegant either, especially since I actually have many more conutries than the two in my toy example above.

What's the proper way to do this with bokeh?

Upvotes: 0

Views: 210

Answers (1)

Tony
Tony

Reputation: 8297

Short and elegant:

from bokeh.palettes import Category10

groups = df.groupby('country')
p = figure(x_axis_type = "datetime")
p.multi_line(xs = [df.date for i, df in groups],
             ys = [df.population for i, df in groups],
             line_color = Category10[10][0: len(groups)],)

More elegant:

from bokeh.palettes import Category10

groups = df.groupby('country')

data = {'date': [], 'population': [], 'legend': []}
for i, df in groups:
    data['date'].append(df.date.tolist())
    data['population'].append(df.population.tolist())
    data['legend'].append(i)
data['color'] = Category10[10][0: len(groups)]

p = figure(x_axis_type = "datetime")
p.multi_line(xs = 'date',
             ys = 'population',
             line_color = 'color',
             legend = 'legend',
             source = data, )

Upvotes: 1

Related Questions