MBSGDRegressor#

class cuml.linear_model.MBSGDRegressor(
*,
loss='squared_loss',
penalty='l2',
alpha=0.0001,
l1_ratio=0.15,
fit_intercept=True,
epochs=1000,
tol=0.001,
shuffle=True,
learning_rate='constant',
eta0=0.001,
power_t=0.5,
batch_size=32,
n_iter_no_change=5,
verbose=False,
output_type=None,
)[source]#

Linear model fitted by minimizing a regularized empirical loss with SGD.

The MBSGD Regressor implementation is experimental and and it uses a different algorithm than sklearn’s SGDClassifier. In order to improve the results obtained from cuML’s MBSGD Regressor:

  • Reduce the batch size

  • Increase the eta0

  • Increase the number of iterations

Since cuML is analyzing the data in batches using a small eta0 might not let the model learn as much as scikit learn does. Furthermore, decreasing the batch size might seen an increase in the time required to fit the model.

Parameters:
loss‘squared_loss’ (default = ‘squared_loss’)

‘squared_loss’ uses linear regression

penalty{‘l1’, ‘l2’, ‘elasticnet’, None} (default = ‘l2’)

The penalty (aka regularization term) to apply.

  • ‘l1’: L1 norm (Lasso) regularization

  • ‘l2’: L2 norm (Ridge) regularization (the default)

  • ‘elasticnet’: Elastic Net regularization, a weighted average of L1 and L2

  • None: no penalty is added

alphafloat (default = 0.0001)

The constant value which decides the degree of regularization

fit_interceptboolean (default = True)

If True, the model tries to correct for the global mean of y. If False, the model expects that you have centered the data.

l1_ratiofloat (default=0.15)

The l1_ratio is used only when penalty = elasticnet. The value for l1_ratio should be 0 <= l1_ratio <= 1. When l1_ratio = 0 then the penalty = 'l2' and if l1_ratio = 1 then penalty = 'l1'

batch_sizeint (default = 32)

It sets the number of samples that will be included in each batch.

epochsint (default = 1000)

The number of times the model should iterate through the entire dataset during training (default = 1000)

tolfloat (default = 1e-3)

The training process will stop if current_loss > previous_loss - tol

shuffleboolean (default = True)

True, shuffles the training data after each epoch False, does not shuffle the training data after each epoch

eta0float (default = 0.001)

Initial learning rate

power_tfloat (default = 0.5)

The exponent used for calculating the invscaling learning rate

learning_rate{‘constant’, ‘invscaling’, ‘adaptive’} (default = ‘constant’)

constant keeps the learning rate constant

adaptive changes the learning rate if the training loss or the validation accuracy does not improve for n_iter_no_change epochs. The old learning rate is generally divided by 5

n_iter_no_changeint (default = 5)

the number of epochs to train without any improvement in the model

verboseint or boolean, default=False

Sets logging level. It must be one of cuml.common.logger.level_*. See Verbosity Levels for more info.

output_type{‘input’, ‘array’, ‘dataframe’, ‘series’, ‘df_obj’, ‘numba’, ‘cupy’, ‘numpy’, ‘cudf’, ‘pandas’}, default=None

Return results and set estimator attributes to the indicated output type. If None, the output type set at the module level (cuml.global_settings.output_type) will be used. See Output Data Type Configuration for more info.

Attributes:
coef_: array, shape=(n_features,)

The model coefficients.

intercept_: float

The independent term. If fit_intercept is False, will be 0.

Methods

fit(X, y, *[, convert_dtype])

Fit the model with X and y.

Notes

For additional docs, see scikitlearn’s SGDRegressor.

Examples

>>> import cupy as cp
>>> import cuml
>>> X = cp.array([[1, 1], [1, 2], [2, 2], [2, 3]], dtype=cp.float32)
>>> y = cp.array([1, 1, 2, 2], dtype=cp.float32)
>>> X_test = cp.asarray([[3, 5], [2, 5]], dtype=cp.float32)
>>> model = cuml.MBSGDRegressor().fit(X, y)
>>> model.predict(X_test)
array([1.5156871, 1.5121976], dtype=float32)
fit(
X,
y,
*,
convert_dtype='deprecated',
) → MBSGDRegressor[source]#

Fit the model with X and y.

Parameters:
Xarray-like (device or host) shape = (n_samples, n_features)

Dense matrix with dtype float32 or float64. Acceptable formats: CUDA array interface compliant objects like CuPy, cuDF DataFrame/Series, NumPy ndarray and Pandas DataFrame/Series.

yarray-like (device or host) shape = (n_samples, 1)

Dense matrix with dtype float32 or float64. Acceptable formats: CUDA array interface compliant objects like CuPy, cuDF DataFrame/Series, NumPy ndarray and Pandas DataFrame/Series.

convert_dtypebool, optional (default = ‘deprecated’)

Deprecated since version 26.08: convert_dtype was deprecated in version 26.08 and will be removed in version 26.10. cuML only copies input arrays when necessary (e.g. to unify dtypes), there is no reason to provide this keyword going forward.

On this page