pylab is a clean way to bulk import a whole slew of helpful functions (the pyplot state machine function, most of numpy) into a single name space. The main reason this exists (to my understanding) is to work with ipython to make a very nice interactive shell which more-or-less replicates MATLAB (to make the transition easier and because it is good for playing around). See pylab.py and matplotlib/pylab.py
At some level, this is purely a matter of taste and depends a bit on what you are doing.
If you are not embedding in a gui (either using a non-interactive backend for bulk scripts or using one of the provided interactive backends) the typical thing to do is
import matplotlib.pyplot as plt
import numpy as np
plt.plot(....)
which doesn't pollute the name space. I prefer this so I can keep track of where stuff came from.
If you use
ipython --pylab
this is equivalent to running
from pylab import *
It is now recommended that for new versions of ipython you use
ipython --matplotlib
which will set up all the proper background details to make the interactive backends to work nicely, but will not bulk import anything. You will need to explicitly import the modules want.
import numpy as np
import matplotlib.pyplot as plt
is a good start.
If you are embedding matplotlib in a gui you don't want to import pyplot as that will start extra gui main loops, and exactly what you should import depends on exactly what you are doing.
From the official documentation, as shown below, the recommendation is to use matplotlib.pyplot. This is not an opinion.
The documentation at Matplotlib, pyplot and pylab: how are they related?, which also describes the difference between pyplot and pylab, states: "Although many examples use pylab, it is no longer recommended.".
Since heavily importing into the global namespace may result in unexpected behavior, the use of pylab is strongly discouraged. Use matplotlib.pyplot instead.