What is the necessity of plt.figure() in matplotlib?

plt.figure(figsize=(10,8))
plt.scatter(df['attacker_size'][df['year'] == 298], # attacker size in year 298 as the y axis df['defender_size'][df['year'] == 298], # the marker as marker='x', # the color color='b', # the alpha alpha=0.7, # with size s = 124, # labelled this label='Year 298')

In the above snippet of code collected from Scatterplot in Matplotlib, what is the necessity of plt.figure()?

2 Answers

The purpose of using plt.figure() is to create a figure object.

The whole figure is regarded as the figure object. It is necessary to explicitly use plt.figure() when we want to tweak the size of the figure and when we want to add multiple Axes objects in a single figure.

# in order to modify the size
fig = plt.figure(figsize=(12,8))
# adding multiple Axes objects
fig, ax_lst = plt.subplots(2, 2) # a figure with a 2x2 grid of Axes

Parts of a Figure

It is not always necessary because a figure is implicitly created when you create a scatter plot; however, in the case you have shown, the figure is being created explicitly using plt.figure so that the figure will be a specific size rather than the default size.

The other option would be to use gcf to get the current figure after creating the scatter plot and set the figure size retrospectively:

# Create scatter plot here
plt.gcf().set_size_inches(10, 8)