def project_rows(self, X, preprocess=True):
''' Project rows on a new subspace formed by a family of eigenvectors. '''
util.verify_dataframe(X)
# Supplementary variables should be preprocessed the same way the active variables were
if preprocess:
X = X.copy(deep=True)
self._preprocess(X)
return X.values @ self.svd.V.T
def __init__(self, dataframe, k, plotter):
if plotter not in 'mpl':
raise ValueError('Unrecognized plotting backend; choose from: mpl')
util.verify_dataframe(dataframe)
self.X = dataframe
self.n, self.p = self.X.shape
# Determine the number of components computed during SVD
self.k = self.p if k == -1 else min(k, self.p)
def __init__(self, dataframe, nbr_components=2, ignored_variable_names=(), scaled=True,
plotter='mpl'):
self.ignored_variable_names = ignored_variable_names
self.scaled = scaled
util.verify_dataframe(dataframe)
self._tidy(dataframe)
super(PCA, self).__init__(dataframe=dataframe, k=nbr_components, plotter=plotter)
self._set_plotter(plotter)
self._preprocess(self.X)
self._compute_svd()
def test_verify_dataframe_failure():
''' Check verify_dataframe raises an exception if the passed argument is not a
pandas.DataFrame. '''
with pytest.raises(ValueError):
util.verify_dataframe((1, 2, 3))
def test_verify_dataframe_success():
''' Check verify_dataframe doesn't raise and exception if the passed argument is a
pandas.DataFrame '''
util.verify_dataframe(pd.DataFrame([1, 2, 3]))