⚙️ Machine Learning without boilerplate. PyCaret does the repetitive work so you can focus on what matters.
PyCaret is an open-source Python library that simplifies the entire ML cycle — from preprocessing to deployment — under a consistent and productive API.
What does PyCaret do?
It’s not an AutoML that decides everything for you. It’s an experimental orchestration layer that accelerates repetitive work:
from pycaret.classification import *
# Complete setup with automatic preprocessing
clf = setup(data=df, target='label')
# Compare all available models
best_model = compare_models()
# Automatic tuning of the best model
tuned_model = tune_model(best_model)
# Save for production
save_model(tuned_model, 'my_model')One line replaces dozens of lines of code.
What’s included?
- Classification, regression, time series, and clustering
- Automatic preprocessing: encoders, scalers, imputation
- Model comparison with configurable metrics
- Built-in hyperparameter tuning
- Custom estimator support and MLOps hooks
Supported models (selection):
- Classification: Logistic Regression, Random Forest, XGBoost, LightGBM, CatBoost, SVM, KNN…
- Regression: same list plus ridge/lasso models
- Time series: ARIMA, Prophet, ExponentialSmoothing, and more
💡 Explanation in a nutshell#
PyCaret positions itself as an “experiment orchestration layer” — it doesn’t make decisions for you but eliminates boilerplate code. It’s inspired by R’s caret package, which popularized the idea of a unified API for testing multiple algorithms. Ideal for the “citizen data scientist” who needs productivity without sacrificing control.
More information at the link 👇
