def fit(
self,
X: numpy.ndarray,
y: numpy.ndarray,
search_spaces: numpy.ndarray,
search_spaces_is_categorical: List[bool],
) -> None:
assert X.shape[0] == y.shape[0]
assert X.shape[1] == search_spaces.shape[0]
assert X.shape[1] == len(search_spaces_is_categorical)
assert search_spaces.shape[1] == 2
encoder = _CategoricalFeaturesOneHotEncoder()
X, search_spaces = encoder.fit_transform(X, search_spaces,
search_spaces_is_categorical)
self._forest.fit(X, y)
self._trees = [
_FanovaTree(e.tree_, search_spaces)
for e in self._forest.estimators_
]
self._features_to_raw_features = encoder.features_to_raw_features
self._variances = {}
if all(tree.variance == 0 for tree in self._trees):
# If all trees have 0 variance, we cannot assess any importances.
# This could occur if for instance `X.shape[0] == 1`.
raise RuntimeError("Encountered zero total variance in all trees.")
def tree() -> _FanovaTree:
sklearn_tree = Mock()
sklearn_tree.n_features = 3
sklearn_tree.node_count = 5
sklearn_tree.feature = [1, 2, -1, -1, -1]
sklearn_tree.children_left = [1, 2, -1, -1, -1]
sklearn_tree.children_right = [4, 3, -1, -1, -1]
sklearn_tree.value = [-1.0, -1.0, 0.1, 0.2, 0.5]
sklearn_tree.threshold = [0.5, 1.5, -1.0, -1.0, -1.0]
search_spaces = numpy.array([[0.0, 1.0], [0.0, 1.0], [0.0, 2.0]])
return _FanovaTree(tree=sklearn_tree, search_spaces=search_spaces)
def fit(
self,
X: numpy.ndarray,
y: numpy.ndarray,
search_spaces: numpy.ndarray,
column_to_encoded_columns: List[numpy.ndarray],
) -> None:
assert X.shape[0] == y.shape[0]
assert X.shape[1] == search_spaces.shape[0]
assert search_spaces.shape[1] == 2
self._forest.fit(X, y)
self._trees = [_FanovaTree(e.tree_, search_spaces) for e in self._forest.estimators_]
self._column_to_encoded_columns = column_to_encoded_columns
self._variances = {}
if all(tree.variance == 0 for tree in self._trees):
# If all trees have 0 variance, we cannot assess any importances.
# This could occur if for instance `X.shape[0] == 1`.
raise RuntimeError("Encountered zero total variance in all trees.")