def __init__(self, **kwargs):
super().__init__(**kwargs)
self.data = []
self.label = []
self.lgbm = lgb.LGBMClassifier()
self._one_hot_encoder = utils.OneHotEncoder()
self.y_shape = None
def fit(self, y):
if not isinstance(y, np.ndarray):
return
if not utils.is_label(y):
return
self.label_encoder = utils.OneHotEncoder()
self.label_encoder.fit_with_labels(y)
def _label_encoding(self, y):
self._label_encoders = []
new_y = []
for temp_y, output_node in zip(y, self.outputs):
hyper_head = output_node
if isinstance(hyper_head, node.Node):
hyper_head = output_node.in_blocks[0]
if (isinstance(hyper_head, head.ClassificationHead)
and utils.is_label(temp_y)):
label_encoder = utils.OneHotEncoder()
label_encoder.fit_with_labels(y)
new_y.append(label_encoder.encode(y))
self._label_encoders.append(label_encoder)
else:
new_y.append(temp_y)
self._label_encoders.append(None)
return new_y
def clear_weights(self):
self.lgbm = lgb.LGBMClassifier()
self._one_hot_encoder = utils.OneHotEncoder()
self._output_shape = None
def __init__(self, seed=None, **kwargs):
super().__init__(seed=seed, **kwargs)
self.lgbm = lgb.LGBMClassifier(random_state=self.seed)
self._one_hot_encoder = utils.OneHotEncoder()
def __init__(self, **kwargs):
super().__init__(**kwargs)
self.lgbm = lgb.LGBMClassifier()
self._one_hot_encoder = utils.OneHotEncoder()