Compute Lasso and optimize lambda without needing cross validation

fast_lasso.py

import numpy as np
import sklearn.linear_model as lm

class FastLasso:
    def __init__(self, verbose=False):
        self.alphas, self.coefs = 2*[None]
        self.score_path_ = None
        self.best_iteration_ = -1
        self.best_score_ = -np.inf
        self.verbose = verbose

    def fit(self, X, y, x_val, y_val):
        """
        Fit lasso without cross validation

        :param X:
        :param y:
        :param eval_set: (x_val, y_val) tuple
        :return:
        """
        from sklearn.metrics import r2_score
        
        # Fit model
        self.alphas, self.coefs, *_ = lm.lasso_path(X, y, verbose=self.verbose)

        # Find optimal coefs
        self.score_path_ = []
        for i in range(self.coefs.shape[1]):
            m_i = self.coefs[:, i]
            score_i = r2_score(y_val, x_val @ m_i)
            if score_i > self.best_score_:
                self.best_iteration_ = i
                self.best_score_ = score_i
            self.score_path_.append(score_i)

        return self

    def predict(self, X):
        return X @ self.coefs[:, self.best_iteration_]
    
    def score(self, X, y):
        from sklearn.metrics import r2_score
        return r2_score(y, self.predict(X))