Developed by Amanda Park & Phil Sattler, feateng is a Python library meant to make feature engineering for data science preprocessing easier. This library gives reproducible functions for the following:
- Data Cleaning
- Reduce levels of categorical variables
- Convert variable types
- Add date column as index
- Remove columns with high NA values
- One-hot encode categorical variables
- Identify outliers
- Transformations
- Normalize data
- Box-cox transformation
- PCA
- Select features by PPS (Predictive Power Score)