FeatWise is a smart, interactive feature engineering generator for data science projects. This tool helps users automatically suggest, explain, and apply new features to their datasets—making the data preparation process faster, more robust, and easier to reproduce.
Build a Python-based web application where users can:
- Upload a CSV dataset.
- Automatically receive suggestions for new features based on their data, including:
- Polynomial features for numeric columns
- Date decompositions (year, month, day, weekday) for date columns
- Text-based features (length, word count, etc.) for text columns
- Interaction features between columns
- Get clear explanations for each suggested feature transformation.
- Select which features to apply using a simple interface.
- Export the transformed dataset and a ready-to-use Python code snippet that reproduces the selected feature engineering steps.
- Python
- Streamlit (or Gradio) for the UI
- pandas, numpy, scikit-learn for data handling and transformations
- Step-by-step implementation of the above features
- Modular code for easy extension (adding new feature types)
- Sample data and demo
- (Optional) Unit tests and deployment instructions
- Advanced feature suggestions using machine learning
- Support for more data formats
- Team collaboration features
- Cloud deployment options
Stay tuned as we build out FeatWise!