Moez Ali, founder and author of PyCaret, has announced a new version of PyCaret. The version 2.3.6. is its biggest release so far in terms of new features, Ali said. PyCaret is an open-source, low-code machine learning library in Python designed to automate machine learning workflows.
The new features include eda, dashboard. convert_model, check_fairness, create_api. create_docker. create_app, and optimize_threshold.
The 2.3.6 version also has made logging level configurable by an environment variable, made the optional path in AWS configurable, fixed TSNE plot with PCA, fixed rendering of streamlit plots, fixed class names in tree plot and fixed NearZeroVariance preprocessor.
The company has also launched new documentation which Ali dubbed The Bible of PyCaret covering all the information, from installation to deployment in one place.
The document is divided into 7 parts detailing information on installation, quickstart, tutorials, modules, preprocessing, functions and release notes. It also covers requirements of dependences, GPU, environment and docker. Additionally, the document has tutorials for regression, NLP, Time series, clustering and classification.
PyCaret is a Python wrapper around several machine learning libraries and frameworks such as scikit-learn, XGBoost, LightGBM, CatBoost, spaCy, Optuna, Hyperopt, Ray etc.
PyCaret is ideal for data scientists looking to increase productivity, Citizen Data Scientists who prefer low code machine learning solutions, Data Science Professionals who want to build rapid prototypes etc.