`scipy.spatial.distance:jensenshannon`

source: /scipy/spatial/distance.py :1205

Signature

   def     jensenshannon (    p    ,    q    ,    base    =  None   ,  * ,     axis    =  0   ,    keepdims    =  False     )

Summary

Compute the Jensen-Shannon distance (metric) between two probability arrays. This is the square root of the Jensen-Shannon divergence.

Extended Summary

The Jensen-Shannon distance between two probability vectors p and q is defined as,

\frac{D ( p ∥ m ) + D ( q ∥ m )}{2}

where $m$ is the pointwise mean of $p$ and $q$ and $D$ is the Kullback-Leibler divergence.

This routine will normalize p and q if they don't sum to 1.0.

Parameters

p : (N,) array_like: left probability vector
q : (N,) array_like: right probability vector
base : double, optional: the base of the logarithm used to compute the output if not given, then the routine uses the default base of scipy.stats.entropy.
axis : int, optional: Axis along which the Jensen-Shannon distances are computed. The default is 0.
versionadded 1.7.0
keepdims : bool, optional: If this is set to True, the reduced axes are left in the result as dimensions with size one. With this option, the result will broadcast correctly against the input array. Default is False.
versionadded 1.7.0

Returns

js : double or ndarray: The Jensen-Shannon distances between p and q along the axis.

Notes

Examples

from scipy.spatial import distance
import numpy as np

✓

distance.jensenshannon([1.0, 0.0, 0.0], [0.0, 1.0, 0.0], 2.0)
distance.jensenshannon([1.0, 0.0], [0.5, 0.5])
distance.jensenshannon([1.0, 0.0, 0.0], [1.0, 0.0, 0.0])

✗

a = np.array([[1, 2, 3, 4],
              [5, 6, 7, 8],
              [9, 10, 11, 12]])
b = np.array([[13, 14, 15, 16],
              [17, 18, 19, 20],
              [21, 22, 23, 24]])

✓

distance.jensenshannon(a, b, axis=0)
distance.jensenshannon(a, b, axis=1)