def merge(X_1, X_2, Y_1, Y_2):
"""Merge two sets into a new set.
Args:
X_1 (np.array): First array of features.
X_2 (np.array): Second array of features.
Y_1 (np.array): First array of labels.
Y_2 (np.array): Second array of labels.
Returns:
A new merged set that was created from `X_1`, `X_2`, `Y_1` and `Y_2`.
"""
logger.info(f'Merging data ...')
# Vertically stacking `X_1` and `X_2`
X = np.vstack((X_1, X_2))
# Horizontally stacking `Y_1` and Y_2`
Y = np.hstack((Y_1, Y_2))
# Checks if `X` and `Y` have the same size
if X.shape[0] != Y.shape[0]:
# If not, raises a SizeError
raise e.SizeError(
f'`(X_1, X_2)` and `(Y_1, Y_2)` should have the same amount of samples')
logger.debug(f'X: {X.shape} | Y: {Y.shape}.')
logger.info('Data merged.')
return X, Y
def test_size_error():
new_exception = exception.SizeError('error')
try:
raise new_exception
except exception.SizeError:
pass
def split_with_index(X, Y, percentage=0.5, random_state=1):
"""Splits data into two new sets.
Args:
X (np.array): Array of features.
Y (np.array): Array of labels.
percentage (float): Percentage of the data that should be in first set.
random_state (int): An integer that fixes the random seed.
Returns:
Two new sets that were created from `X` and `Y`, along their indexes.
"""
logger.info('Splitting data ...')
# Defining a fixed random seed
np.random.seed(random_state)
# Checks if `X` and `Y` have the same size
if X.shape[0] != Y.shape[0]:
# If not, raises a SizeError
raise e.SizeError('`X` and `Y` should have the same amount of samples')
# Gathering the indexes
idx = np.random.permutation(X.shape[0])
# Calculating where sets should be halted
halt = int(len(X) * percentage)
# Dividing the indexes
I_1, I_2 = idx[:halt], idx[halt:]
# Gathering two new sets from `X`
X_1, X_2 = X[I_1, :], X[I_2, :]
# Gathering two new sets from `Y`
Y_1, Y_2 = Y[I_1], Y[I_2]
logger.debug('X_1: %s| X_2: %s | Y_1: %s | Y_2: %s.', X_1.shape, X_2.shape, Y_1.shape, Y_2.shape)
logger.info('Data splitted.')
return X_1, X_2, Y_1, Y_2, I_1, I_2