set backends for different usecases

Always set backends at the top before importing and plotting

set backends for different use cases. Usually notebook for interactive scenario like jupyter notebook or shell. inline for static plots for applications.

Other backend such as qt is to embed with GUIs.

import matplotlib
matplotlib.use('qt5agg') #when wrong argument given, a list of arguments will be given in the error msg.

magic functions in IPython

%matplotlib inline # or notebook
import matplotlib.pyplot as plt

structure

Figure object

This is the object that keeps track all the axes. Typically one Figure object contains many Axes objects. An empty Figure can be created explicitly by plt.figure() and add Axes by add_subplots()

import matplotlib.pyplot as plt

#first way
fig = plt.figure()
ax = fig.subplots() #this add one Axes to the figure
ax.plot(...)

#second way: create one plot
fig, ax = plt.subplots()
ax.plot(...)
#third way
fig, (ax1, ax2) = plt.subplots(1,2)
for each in [ax1, ax2]:
  each.plot(...)
  
#forth way
fig = plt.figure()
gs = fig.add_gridspec(2,2)
ax1 = fig.add_subplot(gs[0,0]) # takes [0,0] position
ax2 = fig.add_subplot(gs[0,1]) # takes [0,1] position
ax3 = fig.add_subplot(gs[1,:]) #takes the whole second row
ax1.plot(...)
ax2.plot(...)
...

Axes object

  • This is what you think of as ‘a plot’, it is the region of the image with the data space.
  • Axes has set methods for ticks, x/y limits, titles, etc.
  • Axes also has 2 or 3 Axis objects which control axies in Axes
  • check api for detailed methods

    Axis object Text Patch etc. Check API when necessary

    Artist object

    The most lowlevel object, all the objects that can be shown in the image is the child of Artist

    twinx() method

    Add a second y axis. Example:

    fig, ax = plt.subplots()
ax.plot(...)
ax2 = ax.twinx() #this will add a second y axis

How to create a figure

create a figure instance holds the plot itself. Then use pyplot to make different kinds of plots. This is a quick way but not for complex plots with several panels.

Note: there is a ‘current’ plot concept in pyplot style. All the operations will be done on ‘current’ plot

pyplot style

### Quick Matlab(pyplot)-style 
plt.figure()
plt.plot()
plt.axis([xmin, xmax, ymin, ymax])
plt.title('')
dir(plt) # see more methods 
### Matlab-stype with subplots
plt.subplot(2,1,1)
plt.plot()
plt.subplot(2,1,2)
plt.plot()

OO-style plot. plt.subplots will create figure object and the axes. axes is a numpy array which is a matrix specifies the location.

fig is the same as instance by plt.figure(). ax holds all the subplot instances. We can switch easily between subplots

### OO-style
fig, ax = plt.subplots(6) # create a flat array of subplots
fig1, ax1 = plt.subplots(3,3) #create a 3x3 array of subplots, in total 9 subplot instances
ax[0].plot() #plot something on ax[0] subplot
ax[0].plot() #plot somethine else on ax[0]
ax[2].plot() #plot on ax[2]
ax[0].set_xlim() #This will not change 'current' plot. see below.
plt.plot()   #this will always plot on 'current' plot, which might be the last plot.
dir(ax) # see more methods
...

conventions of plot method.

fmt = '[marker][line][color]

fmt = '.r--' # point, red dashed line
plt.plot(x, y, fmt)
plt.scatter() #similar to plot(). but it is more refined

Events(interactivity) driven plots.