def test_fisherz_rdm(self):
rdm = np.random.rand(8, 8)
output = fisherz_rdm(rdm)
self.assertEqual(output.shape[0], 8)
rdm = np.random.rand(7, 8)
output = fisherz_rdm(rdm)
self.assertEqual(output, "Invalid input!")
def rdm_correlation_spearman(RDM1,
RDM2,
rescale=False,
permutation=False,
iter=5000):
"""
Calculate the Spearman Correlation between two RDMs
Parameters
----------
RDM1 : array [ncons, ncons]
The RDM 1.
The shape of RDM1 must be [n_cons, n_cons].
n_cons represent the number of conidtions.
RDM2 : array [ncons, ncons].
The RDM 2.
The shape of RDM2 must be [n_cons, n_cons].
n_cons represent the number of conidtions.
fisherz : bool True or False. Default is False.
Do the Fisher-Z transform of the RDMs or not.
rescale : bool True or False. Default is False.
Rescale the values in RDM or not.
Here, the maximum-minimum method is used to rescale the values except for the values on the diagonal.
permutation : bool True or False. Default is False.
Use permutation test or not.
iter : int. Default is 5000.
The times for iteration.
Returns
-------
corr : array [r, p].
The Spearman Correlation result.
The shape of corr is [2], including a r-value and a p-value.
Notes
-----
Don't set both fisherz=True and rescale=True.
"""
if len(np.shape(RDM1)) != 2 or len(
np.shape(RDM2)) != 2 or np.shape(RDM1)[0] != np.shape(
RDM1)[1] or np.shape(RDM2)[0] != np.shape(RDM2)[1]:
return "Invalid input!"
# get number of conditions
cons = np.shape(RDM1)[0]
print(cons)
# calculate the number of value above the diagonal in RDM
n = int(cons * (cons - 1) / 2)
print(np.shape(RDM1))
print(np.shape(RDM2))
if fisherz == True:
RDM1 = fisherz_rdm(RDM1)
RDM2 = fisherz_rdm(RDM2)
if rescale == True:
# flatten the RDM1
vrdm = np.reshape(RDM1, [cons * cons])
# array -> set -> list
svrdm = set(vrdm)
lvrdm = list(svrdm)
lvrdm.sort()
# get max & min
maxvalue = lvrdm[-1]
minvalue = lvrdm[1]
# rescale
if maxvalue != minvalue:
for i in range(cons):
for j in range(cons):
# not on the diagnal
if i != j:
RDM1[i, j] = float(
(RDM1[i, j] - minvalue) / (maxvalue - minvalue))
# flatten the RDM2
vrdm = np.reshape(RDM2, [cons * cons])
# array -> set -> list
svrdm = set(vrdm)
lvrdm = list(svrdm)
lvrdm.sort()
# get max & min
maxvalue = lvrdm[-1]
minvalue = lvrdm[1]
# rescale
if maxvalue != minvalue:
for i in range(cons):
for j in range(cons):
# not on the diagnal
if i != j:
RDM2[i, j] = float(
(RDM2[i, j] - minvalue) / (maxvalue - minvalue))
# initialize two vectors to store the values above the diagnal of two RDMs
v1 = np.zeros([n], dtype=np.float64)
v2 = np.zeros([n], dtype=np.float64)
# assignment
nn = 0
for i in range(cons - 1):
for j in range(cons - 1 - i):
v1[nn] = RDM1[i, i + j + 1]
v2[nn] = RDM2[i, i + j + 1]
nn = nn + 1
print(v1)
print(v2)
# calculate the Spearman Correlation
rp = np.array(spearmanr(v1, v2))
if permutation == True:
rp[1] = permutation_corr(v1, v2, method="spearman", iter=iter)
return rp