Auto-sklearn is an automated machine learning toolkit and a drop-in replacement for a scikit-learn estimator. It automatically searches for the right learning algorithm for a given machine learning task and optimizes its hyperparameters, eliminating the need for manual algorithm selection and hyperparameter tuning.

Pros

• Automates the entire machine learning pipeline, including preprocessing, feature engineering, and model selection
• Saves time and effort in the model development process
• Consistently produces high-quality models, often outperforming manually tuned solutions
• Integrates seamlessly with scikit-learn, making it easy to use for those familiar with the scikit-learn API

Cons

• Can be computationally expensive and time-consuming for large datasets or complex problems
• May not always find the optimal solution, especially for highly specialized or domain-specific problems
• Limited customization options compared to manual tuning
• Requires some understanding of machine learning concepts to interpret and use results effectively

Code Examples

1. Basic usage:

from autosklearn.classification import AutoSklearnClassifier
from sklearn.model_selection import train_test_split
from sklearn.datasets import load_breast_cancer

# Load data and split into train and test sets
X, y = load_breast_cancer(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state=42)

# Create and fit the AutoSklearnClassifier
automl = AutoSklearnClassifier(time_left_for_this_task=120, per_run_time_limit=30)
automl.fit(X_train, y_train)

# Make predictions and print the accuracy score
y_pred = automl.predict(X_test)
print(f"Accuracy score: {automl.score(X_test, y_test)}")

2. Customizing the search space:

from autosklearn.classification import AutoSklearnClassifier
from autosklearn.pipeline.components.classification import ClassifierChoice

# Create a custom classifier choice
custom_classifiers = ClassifierChoice(
    {"random_forest": RandomForestClassifier,
     "extra_trees": ExtraTreesClassifier}
)

# Create AutoSklearnClassifier with custom classifiers
automl = AutoSklearnClassifier(
    time_left_for_this_task=300,
    per_run_time_limit=60,
    include={"classifier": custom_classifiers}
)

# Fit the model (assuming X_train and y_train are defined)
automl.fit(X_train, y_train)

3. Ensemble building:

from autosklearn.classification import AutoSklearnClassifier

# Create AutoSklearnClassifier with ensemble building
automl = AutoSklearnClassifier(
    time_left_for_this_task=300,
    per_run_time_limit=60,
    ensemble_size=50,
    ensemble_nbest=20
)

# Fit the model and print the final ensemble composition
automl.fit(X_train, y_train)
print(automl.show_models())

Getting Started

To get started with auto-sklearn, follow these steps:

1. Install auto-sklearn:

pip install auto-sklearn

2. Import and use AutoSklearnClassifier or AutoSklearnRegressor:

from autosklearn.classification import AutoSklearnClassifier
from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split

# Load data and split into train and test sets
X, y = load_digits(return_X_y=True)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, random_state=42)

# Create and fit the AutoSklearnClassifier
automl = AutoSklearnClassifier(time_left_for_this_task=120, per_run_time_limit=30)
automl.fit(X_train, y_train)

# Evaluate the model
print(f"Accuracy score: {