Rho+ Introduction¶

Rho+ is a library to make creating beautiful, cohesive plots using multiple Python data science tools (matplotlib, seaborn, plotly, bokeh, and altair) easier. It includes a couple different kinds of improvements:

A pair of light and dark themes that cohere with each other and can be used to ensure consistent appearance across Python data visualization tools.
A set of custom colormaps, easy to use in multiple separate libraries, that are designed to work in both light and dark mode.
Several small ergonomic enhancements to plotting functionality that achieve complicated effects without requiring custom code or tweaking.

Let's see it in action!

In [1]:

Copied!





import matplotlib.pyplot as plt
import numpy as np
import seaborn as sns
import rho_plus as rho
import matplotlib.pyplot as plt
import numpy as np
import seaborn as sns
import rho_plus as rho

To show what Rho+ can do, let's make a complicated four-figure plot in Matplotlib.

In [23]:

Copied!





def test_plot(use_rho=False, is_dark=False):
    if use_rho:
        # Use Rho+ styling
        rho.mpl_setup(is_dark, setup=True)
        boxplot = sns.rp_boxplot
        palette = 'rho_heatmap'
    else:
        # use defaults
        plt.rcdefaults()
        boxplot = sns.boxplot
        palette = 'rocket'  # the default

    # Rho+ doesn't pick a font for you: use whatever you like/have installed
    plt.rcParams['font.sans-serif'] = ['Source Han Sans CN']

    fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(2, 2, figsize=(12, 12), gridspec_kw=dict(wspace=0.3, hspace=0.3))

    fig.suptitle('Demo Plot')

    flights = sns.load_dataset('flights')

    boxplot(data=flights, x='month', y='passengers',
            hue='month', legend=None, ax=ax1)
    ax1.set_title('Airline Passengers by Month')
    ax1.set_xlabel('Month')
    ax1.set_ylabel('Airline Passengers')

    if use_rho:
        # move the y-axis label to the top and make it horizontal
        rho.ylabel_top(ax1)

        # make the axes end at the limits of the plot, and show those as ticks
        rho.smart_ticks(ax1)

    xx = np.arange(10) + 1
    ax2.plot(xx, 2 * xx, label='$O(n)$')
    ax2.plot(xx[:5], np.exp(xx[:5] - 0.1), label='Exponential (cut off for space)')
    ax2.plot(xx, xx ** 2, label='$O(n^2)$')
    ax2.plot(xx, xx * np.log(xx) + 5 / (xx + 2), label=r'$O(n \log n)$')
    ax2.plot(xx, np.log(xx), label=r'$O(\log n)$')
    ax2.plot(xx, xx * 0 + 1, label=r'$O(1)$')
    ax2.legend()
    ax2.set_ylabel('Runtime', wrap=True)
    ax2.set_title('Different Kinds of Big-O Growth')

    if use_rho:
        # move the y-axis label to the top and make it horizontal
        rho.ylabel_top(ax2)

        # make the axes end at the limits of the plot, and show those as ticks
        rho.smart_ticks(ax2)

        # use a spline to smooth out the values
        rho.smooth_straight_lines(ax2)

        # add the labels to the ends of the line chart instead of a legend
        rho.line_labels(ax2)

    penguins = sns.load_dataset('penguins')
    im = ax3.scatter(
        data=penguins,
        c='body_mass_g',
        x='bill_depth_mm',
        y='bill_length_mm',
        ec='face',
        cmap='rho_sequential' if use_rho else 'inferno'
    )
    ax3.set_title('Penguin Bills by Body Mass')
    ax3.set_xlabel('Bill Depth (mm)')
    ax3.set_ylabel('Bill Length (mm)')
    fig.colorbar(im, ax=ax3)

    if use_rho:
        rho.smart_ticks(ax3)
        rho.ylabel_top(ax3)

    glue = sns.load_dataset('glue').pivot(index='Model', columns='Task', values='Score')
    sns.heatmap(glue / 100, annot=True, ax=ax4, cmap=palette, norm='logit', cbar=None)

    if use_rho:
        rho.ylabel_top(ax4);

    ax4.set_title('GLUE NLP Benchmarks')

test_plot()
plt.savefig('../../images/default.png')
def test_plot(use_rho=False, is_dark=False):
    if use_rho:
        # Use Rho+ styling
        rho.mpl_setup(is_dark, setup=True)
        boxplot = sns.rp_boxplot
        palette = 'rho_heatmap'
    else:
        # use defaults
        plt.rcdefaults()
        boxplot = sns.boxplot
        palette = 'rocket'  # the default

    # Rho+ doesn't pick a font for you: use whatever you like/have installed
    plt.rcParams['font.sans-serif'] = ['Source Han Sans CN']

    fig, ((ax1, ax2), (ax3, ax4)) = plt.subplots(2, 2, figsize=(12, 12), gridspec_kw=dict(wspace=0.3, hspace=0.3))

    fig.suptitle('Demo Plot')

    flights = sns.load_dataset('flights')

    boxplot(data=flights, x='month', y='passengers',
            hue='month', legend=None, ax=ax1)
    ax1.set_title('Airline Passengers by Month')
    ax1.set_xlabel('Month')
    ax1.set_ylabel('Airline Passengers')

    if use_rho:
        # move the y-axis label to the top and make it horizontal
        rho.ylabel_top(ax1)

        # make the axes end at the limits of the plot, and show those as ticks
        rho.smart_ticks(ax1)

    xx = np.arange(10) + 1
    ax2.plot(xx, 2 * xx, label='$O(n)$')
    ax2.plot(xx[:5], np.exp(xx[:5] - 0.1), label='Exponential (cut off for space)')
    ax2.plot(xx, xx ** 2, label='$O(n^2)$')
    ax2.plot(xx, xx * np.log(xx) + 5 / (xx + 2), label=r'$O(n \log n)$')
    ax2.plot(xx, np.log(xx), label=r'$O(\log n)$')
    ax2.plot(xx, xx * 0 + 1, label=r'$O(1)$')
    ax2.legend()
    ax2.set_ylabel('Runtime', wrap=True)
    ax2.set_title('Different Kinds of Big-O Growth')

    if use_rho:
        # move the y-axis label to the top and make it horizontal
        rho.ylabel_top(ax2)

        # make the axes end at the limits of the plot, and show those as ticks
        rho.smart_ticks(ax2)

        # use a spline to smooth out the values
        rho.smooth_straight_lines(ax2)

        # add the labels to the ends of the line chart instead of a legend
        rho.line_labels(ax2)

    penguins = sns.load_dataset('penguins')
    im = ax3.scatter(
        data=penguins,
        c='body_mass_g',
        x='bill_depth_mm',
        y='bill_length_mm',
        ec='face',
        cmap='rho_sequential' if use_rho else 'inferno'
    )
    ax3.set_title('Penguin Bills by Body Mass')
    ax3.set_xlabel('Bill Depth (mm)')
    ax3.set_ylabel('Bill Length (mm)')
    fig.colorbar(im, ax=ax3)

    if use_rho:
        rho.smart_ticks(ax3)
        rho.ylabel_top(ax3)

    glue = sns.load_dataset('glue').pivot(index='Model', columns='Task', values='Score')
    sns.heatmap(glue / 100, annot=True, ax=ax4, cmap=palette, norm='logit', cbar=None)

    if use_rho:
        rho.ylabel_top(ax4);

    ax4.set_title('GLUE NLP Benchmarks')

test_plot()
plt.savefig('../../images/default.png')

No description has been provided for this image

Now let's see the Rho+ version. Without changing any of your plotting code, the Rho+ theme setup gives you:

A categorical color palette of 12 colors that have been extensively tuned to ensure usability and aesthetic appeal
An uncluttered default theme that matches with the Rho+ themes for other plotting libraries
A paired dark theme that keeps the same color palette if you prefer working in dark mode

By adding a couple function calls and tweaking a couple lines of code, Rho+ can also give you:

Access to a set of sequential colormaps that are optimized for both light and dark mode
A custom box plot theme
Replacing legends in line plots with annotations
Y-axis labels oriented the way most English text is, set at the top
Smoothed-out curves between points in line plots
Axis limits that show the min and max of the data

In [22]:

Copied!

test_plot(use_rho=True)
plt.savefig('../../images/rho-light.png')
test_plot(use_rho=True)
plt.savefig('../../images/rho-light.png')

Changing one line shows what the paired dark theme looks like:

In [21]:

Copied!

test_plot(use_rho=True, is_dark=True)

plt.savefig('../../images/rho-dark.png')
test_plot(use_rho=True, is_dark=True)

plt.savefig('../../images/rho-dark.png')

This is all you need to get started using Rho+! The test_plot code shows the main Python functionality.

For more information on the available color palettes, consult that documentation. If you have improvements to suggest, or things don't work as you expect them to, feel free to file a pull request. Happy graphing!