Implementation:Scikit learn Scikit learn BenchRCV1LogregConvergence

Knowledge Sources	Scikit_learn Scikit-learn Docs
Domains	Machine Learning, Benchmarking
Last Updated	2026-02-08 15:00 GMT

Overview

Concrete tool for benchmarking logistic regression convergence on the RCV1 dataset provided by scikit-learn.

Description

This benchmark script compares the convergence behavior of various logistic regression solvers on the RCV1 text classification dataset. It evaluates scikit-learn's LogisticRegression (with SAG, SAGA, and liblinear solvers), SGDClassifier, and optionally Lightning's implementations. The script uses joblib caching to speed up repeated runs and measures train/test loss and accuracy across different numbers of iterations.

Usage

Use this benchmark to evaluate which logistic regression solver converges fastest on large-scale sparse text classification problems, and to compare the convergence profiles of different optimization algorithms.

Code Reference

Source Location

• Repository: scikit-learn
• File: benchmarks/bench_rcv1_logreg_convergence.py

Signature

def get_loss(w, intercept, myX, myy, C)
def bench_one(name, clf_type, clf_params, n_iter)

from sklearn.linear_model import LogisticRegression, SGDClassifier
from sklearn.datasets import fetch_rcv1

Import

from sklearn.linear_model import LogisticRegression, SGDClassifier
from sklearn.datasets import fetch_rcv1

I/O Contract

Inputs

Name	Type	Required	Description
name	str	Yes	Identifier for the solver being benchmarked
clf_type	class	Yes	Classifier class (LogisticRegression, SGDClassifier, etc.)
clf_params	dict	Yes	Parameters for the classifier constructor
n_iter	int	Yes	Number of iterations to run

Outputs

Name	Type	Description
train_loss	float	Logistic loss on training data
train_score	float	Accuracy on training data
test_score	float	Accuracy on test data
duration	float	Wall clock time of fit
Plot	matplotlib figure	Convergence curves for all solvers

Usage Examples

Basic Usage

from sklearn.datasets import fetch_rcv1
from sklearn.linear_model import LogisticRegression

rcv1 = fetch_rcv1()
X, y = rcv1.data, rcv1.target

clf = LogisticRegression(solver='saga', max_iter=100, random_state=42)
clf.fit(X, y)
print("Score:", clf.score(X, y))

Related Pages

• Principle:Scikit_learn_Scikit_learn_Classification_Prediction