def S2TMB_p(data, target):
# step 1:find the PC set
_, kVar = np.shape(data)
pc_t = []
o_set = [i for i in range(kVar) if i != target]
for x in o_set:
Z = set([target,x]).union(pc_t)
DAG = optimal_network(Z, data)
pc_t = [i for i in range(kVar) if DAG[target, i] == 1 or DAG[i, target] == 1]
# step 2: remove false PC nodes and find spouses
spouses_t = []
varis_set = [i for i in range(kVar) if i != target and i not in pc_t]
for x in varis_set:
Z = set([target, x]).union(set(pc_t)).union(set(spouses_t))
DAG = optimal_network(Z, data)
pc_t = [i for i in range(kVar) if DAG[target, i] == 1 or DAG[i, target] == 1]
sp = [i for i in range(kVar) for j in range(kVar) if i != target and DAG[target, j] == 1 and DAG[i, j] == 1]
spouses_t = set(spouses_t).union(sp)
# step 3: shrink spouses
var_set = spouses_t.copy()
spouses_t = []
for x in var_set:
Z = set([target, x]).union(pc_t).union(spouses_t)
DAG = optimal_network(Z, data)
pc_t = [i for i in range(kVar) if DAG[target, i] == 1 or DAG[i, target] == 1]
spouses_t = [i for i in range(kVar) for j in range(kVar) if i != target and DAG[target, j] == 1 and DAG[i, j] == 1]
MB = list(set(pc_t).union(set(spouses_t)))
return pc_t, MB
def find_potential_neighbors(data, target):
number, kVar = np.shape(data)
O_set = [i for i in range(kVar) if i != target]
ht_set = []
for v in O_set:
Z = list(set(ht_set).union(set([target, v])))
A = optimal_network(Z, data)
ht_set = [
i for i in range(kVar)
if i != target and A[i, target] == 1 or A[target, i] == 1
]
return ht_set
def find_potential_spouses(data, target, ht_set):
_, kVar = np.shape(data)
ht_neighbors = []
for x in ht_set:
vari_set = find_neighbors(data, x)
ht_neighbors = set(ht_neighbors).union(set(vari_set))
O_set = set(ht_neighbors).difference(set(ht_set).union(set([target])))
st_set = []
for v in O_set:
Z = set([target, v]).union(ht_neighbors).union(st_set)
A = optimal_network(Z, data)
st_set = [
i for i in range(kVar) for j in range(kVar)
if i != target and A[target, j] == 1 and A[i, j] == 1
]
return st_set