Getting Started with TPOT

TPOT is an open-source AutoML tool that automates the process of designing machine learning pipelines. Instead of manually selecting models and tuning hyperparameters, you can use TPOT to search for the best combination of data preprocessing steps, models, and settings. TPOT builds on top of scikit-learn and leverages evolutionary algorithms to optimize entire pipelines, saving you time and potentially discovering combinations you might not consider by hand.

from tpot import TPOTClassifier
from sklearn.datasets import load_digits
from sklearn.model_selection import train_test_split

# Load a sample dataset
digits = load_digits()
X_train, X_test, y_train, y_test = train_test_split(
    digits.data, digits.target, train_size=0.75, random_state=42
)

# Initialize TPOTClassifier
tpot = TPOTClassifier(generations=5, population_size=20, verbosity=2, random_state=42, max_time_mins=2)
tpot.fit(X_train, y_train)

# Print the best pipeline found by TPOT
print("Best pipeline:")
print(tpot.fitted_pipeline_)
print("Test accuracy:", tpot.score(X_test, y_test))