### Define the dimension reduction to be used.
# Here we use a classical univariate feature selection based on F-test,
# namely Anova. We set the number of features to be selected to 500
feature_selection = SelectKBest(f_classif, k=500)
### We combine the dimension reduction and the prediction function
anova_svc = Pipeline([('anova', feature_selection), ('svc', clf)])
### Define the cross-validation scheme used for validation.
# Here we use a LeaveOneLabelOut cross-validation on the session, which
# corresponds to a leave-one-session-out
cv = LeaveOneLabelOut(session)
### Compute the prediction accuracy for the different folds (i.e. session)
cv_scores = cross_val_score(anova_svc, X, y, cv=cv, n_jobs=-1, verbose=1)
### Return the corresponding mean prediction accuracy
classification_accuracy = np.mean(cv_scores)
#### Same test using the supervised clustering
#estimator = SVC(kernel='linear', C=1.)
#A = grid_to_graph(n_x=img_shape[0], n_y=img_shape[1], n_z=img_shape[2], mask=mask)
#print "computed connectivity matrix"
#sc = SupervisedClusteringClassifier(estimator=estimator, connectivity=A, n_jobs=1,
# cv=5, n_iterations=50, verbose=1)
#cv_scores = cross_val_score(sc, X, y, cv=cv, n_jobs=4, verbose=1)
#
#sc.fit(X, y)
#computed_coefs = sc.inverse_transform()
# We compute the score for each patient
for i in range(6):
# Using the data corresponding to the patient
X = data[i].data
y = data[i].target
mask = data[i].mask
img_shape = mask.shape
X = X[:, mask!=0]
# Binarizing y to perform classification
y = y.astype(np.bool)
# Computing connectivity matrix
A = grid_to_graph(n_x=img_shape[0], n_y=img_shape[1], n_z=img_shape[2],
mask=mask)
estimator = SVC(kernel='linear', C=1.)
sc = SupervisedClusteringClassifier(estimator=estimator, n_jobs=1,
n_iterations=150, cv=6, connectivity=A, verbose=0)
cv = StratifiedKFold(y, 10)
print "Computing score for the patient %d on 6" % i
cv_scores = cross_val_score(sc, X, y, cv=cv, n_jobs=8, verbose=0)
sc.fit(X, y)
print ". Classification score for patient %d : %f" % (i, np.mean(cv_scores))
print ". Number of parcels : %d" % len(np.unique(sc.labels_))
scores.append(np.mean(cv_scores))
print "===================================="
print "Average score for the whole dataset : %f", np.mean(scores)
mask = data[i].mask
img_shape = mask.shape
X = X[:, mask != 0]
# Binarizing y to perform classification
y = y.astype(np.bool)
# Computing connectivity matrix
A = grid_to_graph(n_x=img_shape[0],
n_y=img_shape[1],
n_z=img_shape[2],
mask=mask)
estimator = SVC(kernel='linear', C=1.)
sc = SupervisedClusteringClassifier(estimator=estimator,
n_jobs=1,
n_iterations=150,
cv=6,
connectivity=A,
verbose=0)
cv = StratifiedKFold(y, 10)
print "Computing score for the patient %d on 6" % i
cv_scores = cross_val_score(sc, X, y, cv=cv, n_jobs=8, verbose=0)
sc.fit(X, y)
print ". Classification score for patient %d : %f" % (i,
np.mean(cv_scores))
print ". Number of parcels : %d" % len(np.unique(sc.labels_))
scores.append(np.mean(cv_scores))
print "===================================="
print "Average score for the whole dataset : %f", np.mean(scores)
