pycaret

What it is and what it does

PyCaret is a low-code machine learning library that automates the end-to-end workflow of building, comparing, and deploying supervised and unsupervised models. It wraps popular libraries like scikit-learn, LightGBM, and statsmodels, reducing hundreds of lines of boilerplate code to a handful of function calls. You initialize a setup with your data and target, then call compare_models() to automatically train and rank candidate algorithms, evaluate_model() to inspect performance, and predict_model() to generate predictions on new data. The library supports classification, regression, time-series forecasting, clustering, and anomaly detection through both a functional API and an object-oriented API.

PyCaret is designed for data scientists who want to prototype quickly, citizen data scientists with moderate technical depth, and ML students. It handles preprocessing, feature engineering, hyperparameter tuning, and model selection internally. The library includes visualization tools (via matplotlib, plotly, yellowbrick) and supports GPU acceleration for tree-based models. With 34 runtime dependencies spanning the scientific Python stack, it trades environment size for convenience—you get a complete ML toolkit in one package.

Use it for

• Rapidly prototype classification or regression models on a new dataset without writing custom preprocessing or cross-validation code.
• Compare multiple algorithms automatically to identify the best performer before investing in tuning.
• Build time-series forecasts or detect anomalies in sensor/log data with minimal boilerplate.
• Train clustering models and visualize results interactively for exploratory data analysis.
• Generate model predictions on hold-out or new data and export trained pipelines for deployment.

Worth the install?

AI-flagged interpretation of the facts on this page. Verify before relying on it.

Install

pycaret on PyPI

Before you install

Low install friction with a pure-wheel distribution. Active maintenance with recent commits and no known vulnerabilities. The 34 runtime dependencies are substantial but standard ML stack (numpy, pandas, scikit-learn, LightGBM, plotly, statsmodels)—expect a moderately sized environment.

Requires Python 3.9 or later. GPU training requires additional setup (e.g., LightGBM GPU libraries) if use_gpu=True is passed.

License in practice

MIT license permits commercial and private use with minimal restrictions. You may use, modify, and distribute PyCaret freely provided you include the license notice.

Quickstart

from pycaret.datasets import get_data
from pycaret.classification import setup, compare_models, predict_model

data = get_data('juice')
s = setup(data, target='Purchase', session_id=123)
best = compare_models()
predictions = predict_model(best, data=data)

Package facts

Evidence: pycaret-3.3.2-py3-none-any.whl

Tags

More Artificial Intelligence packages

See also skforecast · feature-engine · scikit-learn-intelex · azureml-sdk · xgboost-cpu · libcuml-cu12 · sktime · azureml-pipeline · dtreeviz · scikit-base