github.com/redcican/pycaret-streamlit ↗

About redcican/pycaret-streamlit

This is an end-to-end machine learning web application built with Streamlit and PyCaret that handles both classification and regression problems. The application accepts CSV and Excel files for training and provides a complete workflow from data exploration through model deployment, leveraging PyCaret's AutoML capabilities to simplify the machine learning pipeline.

The platform offers extensive preprocessing functionality including data cleaning, feature engineering, feature selection, and unsupervised learning techniques like clustering and outlier removal. For model training, it automatically compares available algorithms, allows training of individual models or ensembles, and supports hyperparameter tuning. Users can visualize results through regression and classification plots as well as SHAP values for model interpretability.

The application supports both online and batch predictions, with the ability to save trained pipelines as pickle files for later use. It can be run locally through Streamlit or deployed via Docker, making it accessible to users without extensive machine learning expertise who need to quickly build and evaluate models.