def fit_pls(X, Y, n_components, scale=True, algorithm="randomized"):
#scaling
print "calculating SVD"
if scale:
X_scaled = zscore(X, axis=0, ddof=1)
Y_scaled = zscore(Y, axis=0, ddof=1)
covariance = np.dot(Y_scaled.T, X_scaled)
else:
covariance = np.dot(Y.T, X)
print np.shape(covariance)
sum_var = covariance
svd = TruncatedSVD(n_components, algorithm)
#computes only the first n_components largest singular values
#produces a low-rank approximation of covariance matrix
Y_saliences, singular_values, X_saliences = svd._fit(covariance)
X_saliences = X_saliences.T
inertia = singular_values.sum()
if scale:
return X_saliences, Y_saliences, singular_values, inertia, X_scaled, Y_scaled, sum_var
else:
return X_saliences, Y_saliences, singular_values, inertia
np.floor(0.8 *
np.shape(brain_data)[0]),
replace=False)
training[training_ind] = True
test = np.invert(training)
#calculate SVD from training data set
X_saliences, Y_saliences, singular_values, inertia, X_scaled, Y_scaled, sum_var = fit_pls(
brain_data[training, :],
Ob[training, :],
n_comp,
scale=True,
algorithm="randomized")
#multiply saliencies with test (leftout) dataset (first scale this one)
brain_test_scaled = zscore(brain_data[test, :], axis=0, ddof=1)
Ob_test_scaled = zscore(Ob[test, :], axis=0, ddof=1)
for comp in np.arange(0, n_comp):
X_saliences_splits[i, comp, :] = X_saliences[:, comp]
Y_saliences_splits[i, comp, :] = Y_saliences[:, comp]
lv_brain = np.matrix(brain_test_scaled) * np.matrix(X_saliences)
lv_ob = np.matrix(Ob_test_scaled) * np.matrix(Y_saliences)
corr_vals[i, comp, 0] = pearsonr(lv_ob[:, comp], lv_brain[:,
comp])[0]
#permute the smaller matrix of the training data and reproject
for j in np.arange(1, perm_n + 1):
Ob_test_scaled_perm = np.random.permutation(Ob_test_scaled)