def fit(self, data_x, data_y, validation_data=None, dataset_split=None,
verbose=None):
"""Fit self-organizing deep learning polynomial neural network
:param data_x : numpy array or sparse matrix of shape [n_samples,n_features]
training data
:param data_y : numpy array of shape [n_samples]
target values
:return an instance of self.
Example of using
----------------
from gmdh import Regressor
model = Regressor()
model.fit(data_x, data_y)
"""
if verbose is not None:
self.verbose = verbose
data_x, data_y = train_preprocessing(data_x, data_y, self.feature_names)
if validation_data is None:
input_train_x, train_y, input_validate_x, validate_y = split_dataset(
data_x, data_y, self.param.seq_type)
input_data_x = data_x
else:
input_validate_x, validate_y = train_preprocessing(
validation_data[0], validation_data[1], self.feature_names)
input_train_x = data_x
train_y = data_y
input_data_x = np.vstack((input_train_x, input_validate_x))
data_x = input_data_x
data_y = np.hstack((train_y, validate_y))
self.n_features = data_x.shape[1]
self.l_count = self.n_features
self.n_train = input_train_x.shape[0]
self.n_validate = input_validate_x.shape[0]
if self.param.normalize:
self.scaler = StandardScaler()
input_train_x = self.scaler.fit_transform(input_train_x)
input_validate_x = self.scaler.transform(input_validate_x)
input_data_x = self.scaler.transform(input_data_x)
train_y, validate_y, data_y = self._preprocess_y(train_y, validate_y, data_y)
fit_data = FitData(input_train_x, train_y,
input_validate_x, validate_y,
data_x, data_y,
input_train_x, input_validate_x, input_data_x)
self._pre_fit_check(train_y, validate_y)
self._fit(fit_data)
return self
def fit(self, data_x, data_y):
"""
Fit multilayered group method of data handling algorithm (model)
Parameters
----------
data_x : numpy array or sparse matrix of shape [n_samples,n_features]
training data
data_y : numpy array of shape [n_samples]
target values
Returns
-------
self : returns an instance of self.
Example of using
----------------
from gmdh import MultilayerGMDH
gmdh = MultilayerGMDH()
gmdh.fit(data_x, data_y)
"""
data_x, data_y, self.data_len = train_preprocessing(
data_x, data_y, self.feature_names)
self.data_y = data_y
if self.param.normalize:
self.scaler = StandardScaler()
self.data_x = self.scaler.fit_transform(data_x)
else:
self.data_x = data_x
self._set_data(self.data_x, self.data_y)
self._pre_fit_check()
self._train_gmdh()
return self