scxpand.linear.linear_params

Parameters for linear classifier training.

Classes

LinearClassifierParam([use_log_transform, ...])

Parameters for linear classifier - aligned with scikit-learn SGDClassifier defaults.

class scxpand.linear.linear_params.LinearClassifierParam(use_log_transform=True, model_type=ModelType.LOGISTIC, alpha=0.0001, penalty='l2', n_epochs=1000, class_weight=None, tol=0.001, l1_ratio=0.15, random_seed=42, batch_size=2048, early_stopping_patience=5, eval_interval=1, train_log_interval=10, sampler_type=SamplerType.RANDOM, mask_rate=0.0, noise_std=0.0, soft_loss_beta=1.0, init_learning_rate=0.0001, learning_rate='optimal', eta0=0.01, power_t=0.5, lr_scheduler_type=LRSchedulerType.CONSTANT_LR, lr_scheduler_config=<factory>, max_iter=1, warm_start=False, average=False, n_iter_no_change=5, validation_fraction=0.1, fit_intercept=True, shuffle=True)

Parameters for linear classifier - aligned with scikit-learn SGDClassifier defaults.

classmethod get_model_type()

Return the model type identifier for this parameter class.

Return type:

ModelType

__init__(use_log_transform=True, model_type=ModelType.LOGISTIC, alpha=0.0001, penalty='l2', n_epochs=1000, class_weight=None, tol=0.001, l1_ratio=0.15, random_seed=42, batch_size=2048, early_stopping_patience=5, eval_interval=1, train_log_interval=10, sampler_type=SamplerType.RANDOM, mask_rate=0.0, noise_std=0.0, soft_loss_beta=1.0, init_learning_rate=0.0001, learning_rate='optimal', eta0=0.01, power_t=0.5, lr_scheduler_type=LRSchedulerType.CONSTANT_LR, lr_scheduler_config=<factory>, max_iter=1, warm_start=False, average=False, n_iter_no_change=5, validation_fraction=0.1, fit_intercept=True, shuffle=True)

get_dataset_params()

Return a DataAugmentParams object with dataset-related parameters.

Return type:

DataAugmentParams

get_lr_scheduler_params()

Return a LRSchedulerParams object with learning rate scheduler parameters.

Return type:

LRSchedulerParams

get_optimizer_params()

Return an OptimizerParams object with optimizer parameters.

Return type:

OptimizerParams

alpha: float = 0.0001

average: bool = False

batch_size: int = 2048

class_weight: str | dict | None = None

early_stopping_patience: int = 5

eta0: float = 0.01

eval_interval: int = 1

fit_intercept: bool = True

init_learning_rate: float = 0.0001

l1_ratio: float = 0.15

learning_rate: str = 'optimal'

lr_scheduler_config: dict

lr_scheduler_type: LRSchedulerType = 'ConstantLR'

mask_rate: float = 0.0

max_iter: int = 1

model_type: ModelType = 'logistic'

n_epochs: int = 1000

n_iter_no_change: int = 5

noise_std: float = 0.0

penalty: str = 'l2'

power_t: float = 0.5

random_seed: int = 42

sampler_type: SamplerType = 'random'

shuffle: bool = True

soft_loss_beta: float | None = 1.0

tol: float = 0.001

train_log_interval: int = 10

use_log_transform: bool = True

validation_fraction: float = 0.1

warm_start: bool = False