def __init__(self, alpha: float = 1.0, max_iter: int = 1000, fit_intercept: bool = True) -> None:

self.alpha: float = alpha # 正則化項の係数

self.max_iter: int = max_iter # 繰り返しの回数

self.fit_intercept: bool = fit_intercept # 切片(i.e., \beta_0)を用いるか

self.coef_ = None # 回帰係数(i.e., \beta)保存用変数

self.intercept_ = None # 切片保存用変数

def _soft_thresholding_operator(self, x: float, lambda_: float) -> float:

if x > 0.0 and lambda_ < abs(x):

return x - lambda_

elif x < 0.0 and lambda_ < abs(x):

return x + lambda_

else:

return 0.0

def fit(self, X: np.ndarray, y: np.ndarray):

if self.fit_intercept:

X = np.column_stack((np.ones(len(X)), X))

beta = np.zeros(X.shape[1])

if self.fit_intercept:

beta[0] = np.sum(y - np.dot(X[:, 1:], beta[1:])) / (X.shape[0])

for iteration in range(self.max_iter):

start = 1 if self.fit_intercept else 0

for j in range(start, len(beta)):

tmp_beta = beta.copy()

tmp_beta[j] = 0.0

r_j = y - np.dot(X, tmp_beta)

arg1 = np.dot(X[:, j], r_j)

arg2 = self.alpha * X.shape[0]

beta[j] = self._soft_thresholding_operator(arg1, arg2) / (X[:, j]**2).sum()

if self.fit_intercept:

beta[0] = np.sum(y - np.dot(X[:, 1:], beta[1:])) / (X.shape[0])

if self.fit_intercept:

self.intercept_ = beta[0]

self.coef_ = beta[1:]

else:

self.coef_ = beta

return self

def predict(self, X: np.ndarray):

y = np.dot(X, self.coef_)

if self.fit_intercept:

y += self.intercept_ * np.ones(len(y))