libpysal.weights.
w_local_cluster
(w)[source]¶Local clustering coefficients for each unit as a node in a graph. [ws]
Parameters: |
|
---|---|
Returns: |
|
Notes
The local clustering coefficient \(c_i\) quantifies how close the neighbors of observation \(i\) are to being a clique:
\[c_i = | \{w_{j,k}\} |/ (k_i(k_i - 1)): j,k \in N_i\]
where \(N_i\) is the set of neighbors to \(i\), \(k_i = |N_i|\) and \(\{w_{j,k}\}\) is the set of non-zero elements of the weights between pairs in \(N_i\). [Watts1998]
Examples
>>> from libpysal.weights import lat2W, w_local_cluster
>>> w = lat2W(3,3, rook=False)
>>> w_local_cluster(w)
array([[1. ],
[0.6 ],
[1. ],
[0.6 ],
[0.42857143],
[0.6 ],
[1. ],
[0.6 ],
[1. ]])