Resources

All of the reading comes from the Tutorial page of matplotlib. I have read the following pages:

  1. Usage Guide
  2. The Lifecycle of a Plot
  3. Customizing Matplotlib with style sheets and rcParams
  4. Artist Tutorial
  5. Legend Guide
  6. Customizing Figure Layouts Using GridSpec and Other Functions

My First Confusion

Which one should I use, matplotlib.pyplot or figure.Figure ?

matplotlib has two interfaces:

  1. Figure being the object-oriented interface
  2. pyplot being the state-based interface.

They are equally powerful and both are illustrated in matplotlib docmentation. But it is better to stick to the one and do no mix them. Some suggestions are:

  1. pyplot behaves like base plot in R and is suggested when generating simple plots in an interactive environment (e.g. Jupyter notebook).
  2. In general, it is suggested to use figure.Figure.

I will focus on using the figure.Figure interface throughout my study notes.

Exploration of Plotting

Some common imports are as follow:

import matplotlib as mpl
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec

I have compiled the following list of examples for generating figures. Each example provides some information on figure customization.

  1. Generating a multi-line plot with a legend and axis labels

  2. Generating a bar plot with customizations of the automatic layout, the figure style, the x-axis, figure saving, and combining multiple visualizations

  3. An example matplotlibrc file with customization settings for all kinds of properties. Typically, we can change them by using the following code. Actually, mpl.rcParams is just a dictionary that controls the default aesthetics. Changing its values will change the default setting.
    • mpl.rcParams['lines.linewidth'] = 2
    • mpl.rcParams.update({'font.size': 22})
    • mpl.rc('lines', linewidth=4, linestyle='-.')
    • mpl.rc('ytick.minor', size=0.33333)
    • ax.plot([1, 2, 3], [1, 3, 2], linewidth=4)

  4. Changing properties of axis containers including ticks and labels

    There are two types of Artists: primitives and containers. The primitives represent the standard graphical objects we want to paint onto our canvas (Line2D, Rectangle, Text, AxesImage, etc.), and the containers are places to put them (Tick, Axis, Axes and Figure).

  5. Fine-tuning legend position using bbox_to_anchor and loc

    Those artists with an empty string as label or with a label starting with “_” will be ignored.

  6. Creating more complex legends with legend handler

  7. Creating complex grid layout with Gridspec and adjusting margins between panels

Performance Considerations

All of the following points and more are available from the Performance section.

  1. Lines: Set rcParams['path.simplify'] and rcParams['path.simplify_threshold'] to speed up line segment redering.
  2. Legends: Specifically provide a location for legends to avoid the expensive default operation of finding the best location.
  3. Presets: Use the fast style for automatic simplification.