def _generate_repeated_sample_indices(random_state, sample_imbalance, y,
verbose):
"""Draw randomly repeated samples, return an array of arrays of indeces to train models on
along with a last sample array as the last one may not be the same size as the others.
"""
class_idxs = _generate_class_indices(y)
class_len = [len(class_idx) for class_idx in class_idxs]
majority_class_idx = np.argmax(class_len)
minority_class_idx = int(not majority_class_idx)
tot_min_samples = class_len[minority_class_idx]
tot_maj_samples = class_len[majority_class_idx]
maj_samples_per_sample = int(tot_min_samples / sample_imbalance)
estimators = math.ceil(tot_maj_samples / maj_samples_per_sample)
maj_indices = class_idxs[majority_class_idx]
# maj_samples is a table of estimator-1 rows by maj_samples_per_sample columns
maj_samples = choice(maj_indices,
size=(estimators - 1, maj_samples_per_sample),
replace=False,
random_state=random_state)
# last_maj_sample is a different length than each row of maj_samples to get every examle into a sample
last_maj_sample = np.setxor1d(maj_samples, maj_indices)
min_indices = class_idxs[minority_class_idx]
samples = np.hstack((maj_samples, np.tile(min_indices,
(estimators - 1, 1))))
last_sample = np.hstack((last_maj_sample, min_indices))
if verbose > 0:
print(
"generating {} samples of indices to use to train multiple estimators, "
"sized {} elements with last being {} elements".format(
len(samples) + 1, len(samples[0]), len(last_sample)))
return samples, last_sample
def fit(self, X, y=None):
"""Fit the model with X.
Samples a couple of random based vectors to approximate a Gaussian
random projection matrix to generate n_components features.
Parameters
----------
X : {array-like}, shape (n_samples, n_features)
Training data, where n_samples in the number of samples
and n_features is the number of features.
Returns
-------
self : object
Returns the transformer.
"""
X = check_array(X)
d_orig = X.shape[1] # Initial number of features
# n_components (self.n) is the final number of features
# times_to_stack_v is the integer division of n by d
# we use times_to_stack_v according to the paper FastFood:
# "When n > d, we replicate (7) for n/d independent random matrices
# Vi, and stack them via Vt = [V_1, V_2, ..., V_(n/d)]t until we have
# enoug dimensions."
self.d, self.n, self.times_to_stack_v = \
Fastfood.enforce_dimensionality_constraints(d_orig,
self.n_components)
self.number_of_features_to_pad_with_zeros = self.d - d_orig
if self.d != d_orig:
warn(
"Dimensionality of the input space as been changed (zero padding) from {} to {}."
.format(d_orig, self.d))
self.G = self.rng.normal(size=(self.times_to_stack_v, self.d))
# G is a random matrix following normal distribution
self.B = choice([-1, 1],
size=(self.times_to_stack_v, self.d),
replace=True,
random_state=self.random_state)
# B is a random matrix of -1 and 1
self.P = np.hstack([(i * self.d) + self.rng.permutation(self.d)
for i in range(self.times_to_stack_v)])
# P is a matrix of size d*n/d = n -> the dimension of the embedding space
# P is for the permutation and respects the V stacks (see FastFood paper)
self.S = np.multiply(
1 / self.l2norm_along_axis1(self.G).reshape((-1, 1)),
chi.rvs(self.d, size=(self.times_to_stack_v, self.d)))
self.H = scipy.linalg.hadamard(self.d)
self.U = self.uniform_vector()
return self
def _generate_sample_indices(random_state,
y,
target_imbalance_ratio,
verbose=0):
"""Private function used to _parallel_build_trees function."""
random_instance = check_random_state(random_state)
class_idxs = _generate_class_indices(y)
class_len = [len(class_idx) for class_idx in class_idxs]
minority_class_idx = np.argmin(class_len)
majority_class_idx = np.argmax(class_len)
min_samples = class_len[minority_class_idx]
maj_samples = int(min_samples / target_imbalance_ratio)
n_samples = min_samples + maj_samples
if verbose > 1:
print(
"len(y):{} target_imbalance_ratio:{} minorities:{} majorities:{} "
"n_samples:{}".format(len(y), target_imbalance_ratio, min_samples,
maj_samples, n_samples))
maj_indices = choice(class_idxs[majority_class_idx],
size=maj_samples,
replace=False,
random_state=random_instance)
min_indices = class_idxs[minority_class_idx]
indices_to_choose_from = np.hstack((min_indices, maj_indices))
if verbose > 99:
print("possible indicies to choose from: {}".format(
indices_to_choose_from))
sample_indices = choice(indices_to_choose_from,
size=n_samples,
replace=True,
random_state=random_instance)
if verbose > 99:
print("chosen indicies: {}".format(sample_indices))
return sample_indices
def test_countvectorizer_vocab_dicts_when_pickling():
rng = np.random.RandomState(0)
vocab_words = np.array(['beer', 'burger', 'celeri', 'coke', 'pizza',
'salad', 'sparkling', 'tomato', 'water'])
for x in range(0, 100):
vocab_dict = dict()
words = choice(vocab_words, size=5, replace=False, random_state=rng)
for y in range(0, 5):
vocab_dict[words[y]] = y
cv = CountVectorizer(vocabulary=vocab_dict)
unpickled_cv = pickle.loads(pickle.dumps(cv))
cv.fit(ALL_FOOD_DOCS)
unpickled_cv.fit(ALL_FOOD_DOCS)
assert_equal(cv.get_feature_names(), unpickled_cv.get_feature_names())
def test_countvectorizer_vocab_sets_when_pickling():
# ensure that vocabulary of type set is coerced to a list to
# preserve iteration ordering after deserialization
rng = np.random.RandomState(0)
vocab_words = np.array(['beer', 'burger', 'celeri', 'coke', 'pizza',
'salad', 'sparkling', 'tomato', 'water'])
for x in range(0, 100):
vocab_set = set(choice(vocab_words, size=5, replace=False,
random_state=rng))
cv = CountVectorizer(vocabulary=vocab_set)
unpickled_cv = pickle.loads(pickle.dumps(cv))
cv.fit(ALL_FOOD_DOCS)
unpickled_cv.fit(ALL_FOOD_DOCS)
assert_equal(cv.get_feature_names(), unpickled_cv.get_feature_names())
def test_countvectorizer_vocab_dicts_when_pickling():
rng = np.random.RandomState(0)
vocab_words = np.array(['beer', 'burger', 'celeri', 'coke', 'pizza',
'salad', 'sparkling', 'tomato', 'water'])
for x in range(0, 100):
vocab_dict = dict()
words = choice(vocab_words, size=5, replace=False, random_state=rng)
for y in range(0, 5):
vocab_dict[words[y]] = y
cv = CountVectorizer(vocabulary=vocab_dict)
unpickled_cv = pickle.loads(pickle.dumps(cv))
cv.fit(ALL_FOOD_DOCS)
unpickled_cv.fit(ALL_FOOD_DOCS)
assert_equal(cv.get_feature_names(), unpickled_cv.get_feature_names())
def test_countvectorizer_vocab_sets_when_pickling():
# ensure that vocabulary of type set is coerced to a list to
# preserve iteration ordering after deserialization
rng = np.random.RandomState(0)
vocab_words = np.array(['beer', 'burger', 'celeri', 'coke', 'pizza',
'salad', 'sparkling', 'tomato', 'water'])
for x in range(0, 100):
vocab_set = set(choice(vocab_words, size=5, replace=False,
random_state=rng))
cv = CountVectorizer(vocabulary=vocab_set)
unpickled_cv = pickle.loads(pickle.dumps(cv))
cv.fit(ALL_FOOD_DOCS)
unpickled_cv.fit(ALL_FOOD_DOCS)
assert_equal(cv.get_feature_names(), unpickled_cv.get_feature_names())
def fit(self, X, y=None):
"""Fit the model with X.
Samples a couple of random based vectors to approximate a Gaussian
random projection matrix to generate n_components features.
Parameters
----------
X : {array-like}, shape (n_samples, n_features)
Training data, where n_samples in the number of samples
and n_features is the number of features.
Returns
-------
self : object
Returns the transformer.
"""
X = check_array(X)
d_orig = X.shape[1]
self.d, self.n, self.times_to_stack_v = \
Fastfood.enforce_dimensionality_constraints(d_orig,
self.n_components)
self.number_of_features_to_pad_with_zeros = self.d - d_orig
self.G = self.rng.normal(size=(self.times_to_stack_v, self.d))
self.B = choice([-1, 1],
size=(self.times_to_stack_v, self.d),
replace=True,
random_state=self.random_state)
self.P = np.hstack([(i * self.d) + self.rng.permutation(self.d)
for i in range(self.times_to_stack_v)])
self.S = np.multiply(
1 / self.l2norm_along_axis1(self.G).reshape((-1, 1)),
chi.rvs(self.d, size=(self.times_to_stack_v, self.d)))
self.U = self.uniform_vector()
return self
def MB_step(X,
x_squared_norms,
centers,
counts,
curr_iter,
old_center_buffer,
compute_squared_diff,
distances,
random_reassign=False,
random_state=None,
reassignment_ratio=.01,
verbose=False,
learn_rate=0.0):
"""Incremental update of the centers for the Minibatch K-Means algorithm.
Parameters
----------
X : array, shape (n_samples, n_features)
The original data array.
x_squared_norms : array, shape (n_samples,)
Squared euclidean norm of each data point.
centers : array, shape (k, n_features)
The cluster centers. This array is MODIFIED IN PLACE
counts : array, shape (k,)
The vector in which we keep track of the numbers of elements in a
cluster. This array is MODIFIED IN PLACE
distances : array, dtype float64, shape (n_samples), optional
If not None, should be a pre-allocated array that will be used to store
the distances of each sample to its closest center.
May not be None when random_reassign is True.
random_state : integer or numpy.RandomState, optional
The generator used to initialize the centers. If an integer is
given, it fixes the seed. Defaults to the global numpy random
number generator.
random_reassign : boolean, optional
If True, centers with very low counts are randomly reassigned
to observations.
reassignment_ratio : float, optional
Control the fraction of the maximum number of counts for a
center to be reassigned. A higher value means that low count
centers are more likely to be reassigned, which means that the
model will take longer to converge, but should converge in a
better clustering.
verbose : bool, optional, default False
Controls the verbosity.
compute_squared_diff : bool
If set to False, the squared diff computation is skipped.
old_center_buffer : int
Copy of old centers for monitoring convergence.
learn_rate: learning rate
Returns
-------
centers:
Updated centers
inertia : float
Sum of distances of samples to their closest cluster center.
squared_diff : numpy array, shape (n_clusters,)
Squared distances between previous and updated cluster centers.
"""
# Perform label assignment to nearest centers
nearest_center, inertia = k_means_._labels_inertia(X,
x_squared_norms,
centers,
distances=distances)
if random_reassign and reassignment_ratio > 0:
random_state = check_random_state(random_state)
# Reassign clusters that have very low counts
to_reassign = counts < reassignment_ratio * counts.max()
# pick at most .5 * batch_size samples as new centers
if to_reassign.sum() > .5 * X.shape[0]:
indices_dont_reassign = np.argsort(counts)[int(.5 * X.shape[0]):]
to_reassign[indices_dont_reassign] = False
n_reassigns = to_reassign.sum()
if n_reassigns:
# Pick new clusters amongst observations with uniform probability
new_centers = choice(X.shape[0],
replace=False,
size=n_reassigns,
random_state=random_state)
if verbose:
print("[MiniBatchKMeans] Reassigning %i cluster centers." %
n_reassigns)
if sp.issparse(X) and not sp.issparse(centers):
assign_rows_csr(X, astype(new_centers, np.intp),
astype(np.where(to_reassign)[0], np.intp),
centers)
else:
centers[to_reassign] = X[new_centers]
# reset counts of reassigned centers, but don't reset them too small
# to avoid instant reassignment. This is a pretty dirty hack as it
# also modifies the learning rates.
counts[to_reassign] = np.min(counts[~to_reassign])
squared_diff = 0.0
## implementation for the sparse CSR representation completely written in
# cython
if sp.issparse(X):
if compute_squared_diff:
old_center_buffer = centers
#rand_vec = make_rand_vector(X.shape[1])
#learn_rate = 0.0
centers = _MB_step._mini_batch_update_csr(X, x_squared_norms, centers,
counts, nearest_center,
old_center_buffer,
compute_squared_diff,
curr_iter, learn_rate)
if compute_squared_diff:
diff = centers - old_center_buffer
squared_diff = row_norms(diff, squared=True).sum()
return centers, squared_diff, inertia
## dense variant in mostly numpy (not as memory efficient though)
k = centers.shape[0]
for center_idx in range(k):
# find points from minibatch that are assigned to this center
center_mask = nearest_center == center_idx
old_count = counts[center_idx]
this_count = center_mask.sum()
counts[center_idx] += this_count # update counts
if this_count > 0:
new_count = counts[center_idx]
if compute_squared_diff:
old_center_buffer[:] = centers[center_idx]
# inplace remove previous count scaling
#centers[center_idx] *= counts[center_idx]
# inplace sum with new points members of this cluster
#centers[center_idx] += np.sum(X[center_mask], axis=0)
# update the count statistics for this center
#counts[center_idx] += count
# inplace rescale to compute mean of all points (old and new)
#centers[center_idx] /= counts[center_idx]
new_center = np.sum(X[center_mask], axis=0)
if learn_rate == 0.0:
learn_rate = (new_count - old_count) / float(new_count)
centers[center_idx] = centers[center_idx] + learn_rate * (
new_center / (new_count - old_count) - centers[center_idx])
# update the squared diff if necessary
if compute_squared_diff:
diff = centers[center_idx].ravel() - old_center_buffer.ravel()
squared_diff += np.dot(diff, diff)
return centers, squared_diff, inertia
def test_non_p_array():
keeper = choice(np.array(1), size=())
assert_equal(keeper[()], 0)
def test_returns_1d_scalar_object_based_off_distribution():
keeper = choice(3, replace=False, p=np.array([0, 0, 1.0]))
assert_equal(keeper, 2)
