"""
Implementation of LASSO regression model.
"""
from tno.mpc.mpyc.secure_learning.models.secure_linear import Linear
from tno.mpc.mpyc.secure_learning.models.secure_model import PenaltyTypes, SolverTypes
[docs]
class Lasso(Linear):
r"""
Solver for LASSO regression. Optimizes a model with objective function
$$\frac{1}{2n_{\textrm{samples}}} \times ||y - X_times_w||^2_2 + \alpha_1 ||w||_1$$
"""
name = "LASSO regression"
[docs]
def __init__(
self,
solver_type: SolverTypes = SolverTypes.GD,
alpha: float = 1,
) -> None:
"""
Constructor method.
:param solver_type: Solver type to use (e.g. Gradient Descent aka GD)
:param alpha: Regularization parameter
"""
super().__init__(solver_type, penalty=PenaltyTypes.L1, alpha=alpha)