Solution 1:

np.random.permutation has two differences from np.random.shuffle:

  • if passed an array, it will return a shuffled copy of the array; np.random.shuffle shuffles the array inplace
  • if passed an integer, it will return a shuffled range i.e. np.random.shuffle(np.arange(n))

If x is an integer, randomly permute np.arange(x). If x is an array, make a copy and shuffle the elements randomly.

The source code might help to understand this:

3280        def permutation(self, object x):
...
3307            if isinstance(x, (int, np.integer)):
3308                arr = np.arange(x)
3309            else:
3310                arr = np.array(x)
3311            self.shuffle(arr)
3312            return arr

Solution 2:

Adding on to what @ecatmur said, np.random.permutation is useful when you need to shuffle ordered pairs, especially for classification:

from np.random import permutation
from sklearn.datasets import load_iris
iris = load_iris()
X = iris.data
y = iris.target

# Data is currently unshuffled; we should shuffle 
# each X[i] with its corresponding y[i]
perm = permutation(len(X))
X = X[perm]
y = y[perm]