ColorScale

The colors for the ColorScale can be defined one of two ways:

Manually, by setting the scale's colors attribute to a list of css colors. They can be either:
- html colors (link) 'white'
- hex '#000000'
- rgb 'rgb(0, 0, 0)'.

col_sc = ColorScale(colors=['yellow', 'red'])

Using one of bqplot's color-schemes. As of now we support all the colorbrewer schemes (link), as well as the matplotlib schemes 'viridis', 'magma', 'inferno' and 'plasma'.
```
col_sc = ColorScale(scheme=['viridis'])
```

The color scale then linearly interpolates between its colors.

ColorAxis

A ColorAxis, like other Axis types, takes a color scale as input. It can then be displayed in a Figure.

ax_col = ColorAxis(scale=col_sc)



In [ ]:

    
import numpy as np

import bqplot.pyplot as plt
from bqplot import ColorScale, DateColorScale, OrdinalColorScale, ColorAxis



In [ ]:

    
# setup data for plotting
np.random.seed(0)
n = 100
x_data = range(n)
y_data = np.cumsum(np.random.randn(n) * 100.0)



In [ ]:

    
def create_fig(color_scale, color_data, fig_margin=None):
    # allow some margin on right for color bar
    if fig_margin is None:
        fig_margin = dict(top=50, bottom=70, left=50, right=100)
    fig = plt.figure(title='Up and Down', fig_margin=fig_margin)
    
    # setup color scale
    plt.scales(scales={'color': color_scale})
    
    # show color bar on right
    axes_options = {'color': {'orientation': 'vertical', 'side': 'right'}}
    scat = plt.scatter(x_data, y_data, color=color_data, stroke='black', axes_options=axes_options)
    return fig



In [ ]:

    
fig = create_fig(ColorScale(), y_data)
fig

Attributes

ColorScales share attributes with the other Scale types:

Their domain can be manually constrained with the min and max attributes
They can be inversed by setting the reverse attribute to True



In [ ]:

    
color_scale = fig.marks[0].scales['color']
color_scale.min = 0



In [ ]:

    
color_scale.reverse = True

Mid

In addition they also have a mid attribute, a value that will be mapped to the middle color. This is especially suited to diverging color schemes.



In [ ]:

    
color_scale.min = None
color_scale.mid = 0

DateColorScale

The DateColorScale is a color scale for dates. It works in every way like the regular ColorScale, except that its min, mid and max attributes — if defined — must be date elements (datetime, numpy or pandas).



In [ ]:

    
import pandas as pd

fig_margin = dict(top=50, bottom=70, left=50, right=200)
date_col_sc = DateColorScale()
dates = pd.date_range(start='2015-01-01', periods=n)
create_fig(date_col_sc, dates, fig_margin=fig_margin)



In [ ]:

    
date_col_sc.min = pd.datetime(2016, 2, 28)

Color Schemes

Use the following widgets to browse through the available color schemes



In [ ]:

    
from bqplot.market_map import MarketMap

from ipywidgets import IntSlider, SelectionSlider, Dropdown
from ipywidgets import VBox, HBox, Layout
from traitlets import link

Diverging schemes



In [ ]:

    
div_schemes = [
    'Spectral',
    'RdYlGn',
    'RdBu',
    'PiYG',
    'PRGn',
    'RdYlBu',
    'BrBG',
    'RdGy',
    'PuOr',
]



In [ ]:

    
def scheme_inspector(fig, schemes, title=''):
    '''
    Takes a Figure and a list of schemes and returns the Figure along with
    dropdown to go through the different schemes
    '''
    
    # Get the color scale
    col_sc = fig.marks[0].scales['color']
    
    # Create the widgets to select the colorscheme
    scheme_dd = Dropdown(description='Scheme',
                         options=schemes)
    
    def update_scheme(*args):
        col_sc.scheme = scheme_dd.value

    scheme_dd.observe(update_scheme, 'value')    
    update_scheme()
    return VBox([scheme_dd, fig])

scheme_inspector(create_fig(ColorScale(), y_data), div_schemes)

Non-diverging schemes



In [ ]:

    
lin_schemes = [
    'OrRd',
    'PuBu',
    'BuPu',
    'Oranges',
    'BuGn',
    'YlOrBr',
    'YlGn',
    'Reds',
    'RdPu',
    'Greens',
    'YlGnBu',
    'Purples',
    'GnBu',
    'Greys',
    'YlOrRd',
    'PuRd',
    'Blues',
    'PuBuGn',
    'viridis',
    'plasma',
    'inferno',
    'magma'
]



In [ ]:

    
scheme_inspector(create_fig(ColorScale(), y_data), lin_schemes, title='Non-diverging schemes')

OrdinalColorScale

The OrdinalColorScale is a color scale for categorical data, i.e. data that does not have an intrinsic order.

The scale colors may be specified by the user, or chosen from a set scheme. As of now, the supported color schemes are the colorbrewer categorical schemes, listed here along with their maximum number of colors.



In [ ]:

    
ord_schemes = {
    'Set2': 8,
    'Accent': 8,
    'Set1': 9,
    'Set3': 12,
    'Dark2': 8,
    'Paired': 12,
    'Pastel2': 8,
    'Pastel1': 9,
}



In [ ]:

    
def partition(array, n_groups):
    n_elements = len(array)
    if n_groups > n_elements:
        return np.arange(n_elements)
    n_per_group = n_elements // n_groups + (n_elements % n_groups > 0)
    return np.tile(range(1, n_groups+1), n_per_group)[:n_elements]



In [ ]:

    
# Define the control widgets
n_groups_slider = IntSlider(description='n colors', min=3)
scheme_dd = Dropdown(description='Scheme',
                     options=ord_schemes)
    
def update_scheme(*args):
    col_sc.scheme = scheme_dd.label
    ax_c.label = scheme_dd.label
    n_groups_slider.max = scheme_dd.value

def update_categories(*args):
    groups = partition(names, n_groups_slider.value)
    market_map.color = groups
    market_map.groups = groups

n_groups_slider.observe(update_categories, 'value')
scheme_dd.observe(update_scheme)

# Define the bqplot marketmap
names = range(100)

col_sc = OrdinalColorScale()
ax_c = ColorAxis(scale=col_sc)

market_map = MarketMap(names=names,
                       display_text=['' for _ in names],
                       scales={'color': col_sc}, axes=[ax_c],
                       layout=Layout(min_width='800px', min_height='600px'))
update_scheme()
update_categories()

VBox([HBox([scheme_dd, n_groups_slider]), market_map])