population.py

''' A chart that uses the population dataset of the globe.
This example shows the ability of Bokeh to use two different kinds of graph into
one file and sorting them by having different colors and lines.

'''
from math import pi

from bokeh.io import curdoc
from bokeh.layouts import column, row
from bokeh.models import (ColumnDataSource, CustomJSTickFormatter,
                          CustomJSTransform, Select)
from bokeh.plotting import figure
from bokeh.sampledata.population import data as df
from bokeh.transform import factor_cmap, transform

years = [str(x) for x in sorted(df.Year.unique())]
locations = sorted(df.Location.unique())
groups = list(df.AgeGrp.unique())

# pyramid plot of age groups by gender

ages = ColumnDataSource(data=dict(AgeGrp=[], Sex=[], Value=[]))

gender_transform = CustomJSTransform(args=dict(source=ages), func="", v_func="""
    const val = new Float64Array(xs.length)
    for (let i = 0; i < xs.length; i++) {
        if (source.data['Sex'][i] == 'Male')
            val[i] = -xs[i]
        else
            val[i] = xs[i]
    }
    return val
""")

pyramid = figure(width=600, height=500, toolbar_location=None, y_range=groups,
                 title="Population Breakdown by Age Group and Gender",
                 x_axis_label="Population (Millions)",y_axis_label="Age Group")
pyramid.hbar(y="AgeGrp", height=1, right=transform('Value', gender_transform),
             source=ages, legend_field="Sex", line_color="white",
             fill_color=factor_cmap('Sex', palette=["#3B8686", "#CFF09E"], factors=["Male", "Female"]))

pyramid.ygrid.grid_line_color = None
pyramid.xaxis.formatter = CustomJSTickFormatter(code="""
    return (Math.abs(tick) / 1e6) + " M"
""")

# line plot of known and predicted population

known = ColumnDataSource(data=dict(x=[], y=[]))
predicted = ColumnDataSource(data=dict(x=[], y=[]))

population = figure(width=600, height=180, toolbar_location=None,
                    title="Total Population by Year",
                    x_axis_label="Year",y_axis_label="Population")
population.line("x", "y", color="violet", line_width=2, source=known, legend_label="known")
population.line("x", "y", color="violet", line_width=2, line_dash="dashed", source=predicted, legend_label="predicted")

population.xaxis.major_label_orientation = pi/4
population.xgrid.grid_line_color = None
population.legend.location = "center_right"
population.x_range.end = 2150
population.yaxis.minor_tick_line_color = None
population.yaxis.formatter = CustomJSTickFormatter(code="""
    return (Math.abs(tick) / 1e9) + " B"
""")

# Controls and callbacks

year = Select(title="Year:", value="2010", options=years)
location = Select(title="Location:", value="World", options=locations)

def update():
    ages.data = df[(df.Location == location.value) & (df.Year == int(year.value))]

    pop = df[df.Location == location.value].groupby(df.Year).Value.sum()
    new_known = pop[pop.index <= 2010]
    new_predicted = pop[pop.index >= 2010]
    known.data = dict(x=new_known.index.map(str), y=new_known.values)
    predicted.data = dict(x=new_predicted.index.map(str), y=new_predicted.values)

year.on_change('value', lambda attr, old, new: update())
location.on_change('value', lambda attr, old, new: update())

update()

controls = column(year, location, width=300)
curdoc().add_root(row(column(pyramid, population), controls))