`cda_fast` benchmark

benchmark.py

import warnings
from sklearn.exceptions import ConvergenceWarning
warnings.filterwarnings("ignore", category=ConvergenceWarning)

from sklearn.linear_model import Lasso
import numpy as np
import time

clf = Lasso(max_iter=200)
n_samples = 500000
n_features = 200
X = np.random.rand(n_samples, n_features)
y = np.ones((n_samples, 1))
y = y.ravel()
def _time_fit(i):
    start_time = time.time()
    clf.fit(X, y)
    return time.time() - start_time

iter = 10
times = np.array([_time_fit(i) for i in range(iter)])
print(f"{times=}\n{times.mean():.2f}+-{times.std():.2f}")