What is PyCaret?
PyCaret is an open source low code end-to-end machine learning library in Python. Its primary objective is to reduce the cycle time of hypothesis to insights and make data scientists more productive in their experiments. It does this by providing a high-level API which is sophisticated yet easy to use for data scientists and analysts who seek to perform iterative, end-to-end data science experiments in a very efficient way. Through the use of PyCaret, the amount of time spent in coding experiments reduce drastically (up to 20 folds).
The design and simplicity of PyCaret is inspired by the emerging role of citizen data scientists. In comparison to other open source machine learning libraries, PyCaret is a low-code solution which is simple in design and easy to use. The architecture of PyCaret is deployment ready which means all steps and dependencies in an experiment are automatically orchestrated and saved into a pipeline that can be deployed into production or could be transferred into another environment to run at scale.
As of the first public release, PyCaret supports Classification and Regression in supervised learning and Anomaly Detection, Clustering, Natural Language Processing and Association Rule Mining in unsupervised learning. It has over 70 ready-to-use open source algorithms and over 25 pre-processing techniques that are fully orchestrated. PyCaret supports automatic hyperparameter tuning, automatic feature engineering and feature selection. It also has rich analytical capabilities with over 40 interactive visualizations to analyze machine learning models. Future releases will include Time Series Modeling, Recommender System and Deep Learning modules.
Who should use PyCaret?
PyCaret is free and open source library which is easy to install and can be setup either locally or on any cloud service within minutes. The licensing agreement also allows for the commercial use of the software. While there is no limitation of use, the ideal target audience is as follows:
- Citizen data scientists and analysts who wants to easily implement end-to-end data science projects in a low-code environment.
- Data scientists who want to increase the productivity and efficiency of their experiments.
- Data science students and analytics practitioners with no prior background in coding.
- Small to midsize companies looking to implement data science projects without committing significant amounts of resources.
PyCaret in Analytical Ecosystem
PyCaret integrates seamlessly with tools and platforms that support python such as Microsoft Power BI, Tableau, Alteryx and KNIME to name a few. This gives immense power to users who can now integrate PyCaret into their existing workflows and add a layer of Machine Learning to their analytical applications very easily at no cost.
Want to get started early?
You can become an early adopter and download the pre-release build (pycaret 0.0.60 at the time of this post) using pip installer. Builds are refreshed nightly.
pip install pycaretGetting Started Tutorials
- Getting Started — Binary Classification
- Getting Started — Multiclass Classification
- Getting Started — Regression
- Getting Started — Clustering
- Getting Started — Anomaly Detection
- Getting Started — Natural Language Processing
- Getting Started — Association Rule Mining
Tutorials are updated frequently.
Other Resources
Github Page: http://www.github.com/pycaret/pycaret
Official Website: http://www.pycaret.org (work in progress)
Linkedin: https://www.linkedin.com/company/pycaret/
Comments / Feedback
Comments and feedback are welcome at moez@pycaret.org
