def pn_select(scores, blacklist):
pnl = [pn for pn in scores.iterkeys() if pn not in blacklist]
if len(pnl) == 0:
return None # no suitable product node found
pnsl = [scores[pn] for pn in pnl]
pni = argmin_fair(pnsl)
return pnl[pni]
def ham_cluster(data, k):
m, n = data.shape
assert 1 <= k <= m
pts = range(m)
p = random.choice(pts)
pts.remove(p)
centers = [p]
distances = [[hamming(data[i], data[p])] for i in range(m)]
for i in range(k-1):
pi = argmax_fair([min(distances[q]) for q in pts])
p = pts[pi]
for q in pts:
distances[q].append( hamming(data[q], data[p]) )
centers.append(p)
pts.remove(p)
a = [argmin_fair(distances[q]) for q in range(m)] # assignments
return np.array(a)
