AutoML with auto_sklearn2

AutoML systems such as auto_sklearn2 automate the most time-consuming and technically challenging parts of the machine learning workflow. When you use auto_sklearn2, it automatically performs:

• Data preprocessing;
• Feature engineering;
• Model selection;
• Hyperparameter tuning.

You do not need to manually test different algorithms or tune their parameters one at a time. auto_sklearn2 explores a wide range of models and configurations for you, using advanced optimization techniques to search for the best-performing pipeline on your dataset.

The system also includes built-in ensembling, so your final model benefits from the strengths of several algorithms combined together.

from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split
from auto_sklearn2.classification import AutoMLClassifier

# Load a sample dataset
X, y = load_digits(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=42)

# Initialize and fit AutoMLClassifier from auto_sklearn2
automl = AutoMLClassifier(
    time_left_for_this_task=120,   # total runtime for the search
    per_run_time_limit=30,         # limit per model evaluation
    ensemble_size=50,
    seed=42
)
automl.fit(X_train, y_train)

# Display the leaderboard of discovered models
print(automl.leaderboard(X_test, y_test, detailed=True))