def transform(self, values):
"""Transform values and return the resulting feature matrix
Parameters
----------
values : array-like, [n_samples]
Returns
-------
X : sparse matrix, shape [n_samples, n_features]
Raises
------
NotFittedError
If the vectorizer has not yet been fitted.
"""
if not hasattr(self, 'feature_names_'):
raise utils.NotFittedError('Vectorizer has not yet been fitted')
if self._lowercase:
values = [value.lower() for value in values]
values = np.asarray(values).reshape(-1, 1)
return self._vectorizer.transform(values)
def transform(self, values):
"""Transform values and return the resulting feature matrix
Parameters
----------
values : array-like, [n_samples]
Returns
-------
X : sparse matrix, [n_samples, n_features]
Raises
------
NotFittedError
If the vectorizer has not yet been fitted.
"""
if not hasattr(self, 'feature_names_'):
raise utils.NotFittedError('Vectorizer has not yet been fitted')
values = np.asarray(values)
indices = np.digitize(values, self._bin_edges)
ones = np.ones(indices.shape[0], dtype=bool)
return sp.coo_matrix(
(ones, (np.arange(indices.shape[0]), indices)),
shape=(indices.shape[0], len(self.feature_names_)),
dtype=bool)
def transform(self, values):
"""Transform values to feature matrix
Parameters
----------
values : array-like, [n_samples]
Strings for transforming.
Returns
-------
X : sparse matrix, shape [n_samples, n_features]
Feature matrix.
Raises
------
NotFittedError
If the vectorizer has not yet been fitted.
ValueError
If `values` is not a one-dimensional array.
"""
if not hasattr(self, 'feature_names_'):
raise utils.NotFittedError('Vectorizer has be yet been fitted')
if self._categorical:
values = [value.lower() for value in values]
return self._vectorizer.transform(values).astype(bool)
else:
return self._vectorizer.transform(values)
def transform(self, values):
"""Transform booleans to feature matrix
Parameters
----------
values : array-like, [n_samples]
Booleans for transforming.
Returns
-------
X : ndarray, [n_samples, 1]
Feature matrix.
Raises
------
NotFittedError
If the vectorizer has not yet been fitted.
ValueError
If `values` is not a one-dimensional array.
"""
if not hasattr(self, 'feature_names_'):
raise utils.NotFittedError('Vectorizer has be yet been fitted')
values = np.asarray(values, dtype=bool)
if values.ndim != 1:
raise ValueError(
'values must be a one dimensional array, not with shape {}'.
format(values.shape))
return np.expand_dims(values, 1)
def transform(self, values):
"""Transform numbers to feature matrix
Parameters
----------
values : array-like, [n_samples]
Numbers for transforming.
Returns
-------
X : sparse matrix, [n_samples, n_features]
Feature matrix.
Raises
------
NotFittedError
If the vectorizer has not yet been fitted.
ValueError
If `values` is not a one-dimensional array.
"""
if not hasattr(self, 'feature_names_'):
raise utils.NotFittedError('Vectorizer has be yet been fitted')
values = np.asarray(values)
if values.ndim != 1:
raise ValueError(
'values must be a one dimensional array, not with shape {}'.
format(values.shape))
indices = np.digitize(values, self._bin_edges)
ones = np.ones(indices.shape[0], dtype=bool)
return sp.coo_matrix(
(ones, (np.arange(indices.shape[0]), indices)),
shape=(indices.shape[0], len(self.feature_names_)),
dtype=bool)