def w_knn(wed, n, cost=None, ids=None):
'''
Generate w Weights object based on the k-nearest
network neighbors.
Parameters
----------
wed: PySAL Winged Edged Data Structure
n: integer number of neighbors for each node
cost: defaults to length, can be any cost dictionary
Returns
-------
ps.W(neighbors): PySAL Weights Dict
'''
if cost is None:
cost = edge_length(wed)
if ids:
ids = np.array(ids)
else:
ids = np.arange(len(wed.node_list))
neighbors = {}
for node in wed.node_list:
neighbors[node] = knn_distance(wed, cost, node, n=n)
return ps.W(neighbors, id_order=ids)
def w_distance(wed, threshold, cost=None, alpha=-1.0, binary=True, ids=None):
'''
Generate a Weights object based on a threshold
distance using a WED
Parameters
----------
wed: PySAL Winged Edged Data Structure
distance: float network threshold distance for neighbor membership
cost: defaults to length, can be any cost dicationary {(edge): cost}
Returns
-------
ps.W(neighbors): PySAL Weights Dict
'''
if cost is None:
cost = edge_length(wed)
if ids:
ids = np.array(ids)
else:
ids = np.arange(len(wed.node_list))
neighbors = {}
if binary is True:
for node in wed.node_list:
near, pred = threshold_distance(wed, cost, node, threshold)
neighbors[ids[node]] = near
return ps.W(neighbors, None, ids)
elif binary is False:
weights = {}
for node in wed.node_list:
wt = []
near, pred = threshold_distance(wed, cost, node, threshold)
near.remove(node)
neighbors[ids[node]] = near
for end in near:
path = [end]
previous = pred[end]
while previous != node:
path.append(previous)
end = previous
previous = pred[end]
path.append(node)
cum_cost = 0
for p in range(len(path) - 1):
cum_cost += cost[(path[p], path[p + 1])]
wt.append(cum_cost ** alpha)
weights[ids[node]] = wt
return ps.W(neighbors, weights, ids)
def edge_length(self):
return util.edge_length(self)
def edge_length(self):
return util.edge_length(self)
