def gen_data_xvalidate_objective(estimator, N, d, num_folds=5):
assert_positive_int(num_folds)
O = []
for i in range(num_folds):
x_train = np.random.randn(N, d) # not a proper CV
x_test = np.random.randn(N, d)
try:
estimator.fit(x_train)
except Exception, e:
print(N, d, e)
i -= 1
O.append(estimator.objective(x_test))
def xvalidate_objective(self, X, num_folds=5, num_repetitions=1):
assert_array_shape(X, ndim=2, dims={1: self.D})
assert_positive_int(num_folds)
assert_positive_int(num_repetitions)
O = np.zeros((num_repetitions, num_folds))
for i in range(num_repetitions):
xval = XVal(N=len(X), num_folds=num_folds)
for j, (train, test) in enumerate(xval):
self.fit(X[train])
O[i, j] = self.objective(X[test])
return O
def xvalidate_objective(self, X, num_folds=5, num_repetitions=1):
assert_array_shape(X, ndim=2, dims={1: self.D})
assert_positive_int(num_folds)
assert_positive_int(num_repetitions)
O = np.zeros((num_repetitions, num_folds))
for i in range(num_repetitions):
xval = XVal(N=len(X), num_folds=num_folds)
for j, (train, test) in enumerate(xval):
self.fit(X[train])
O[i, j] = self.objective(X[test])
return O