GitHub - kvh/ramp: Rapid Machine Learning Prototyping in Python

Abstraction: Pandas-based Python library for declarative ML feature and model exploration

Key points:

• Light-weight pandas-based ML framework; pluggable with scikit-learn, rpy2, and others
• Declarative syntax for chaining feature transforms: e.g. Normalize(Log('x')), Interactions([...])
• Supports PCA-based dimension reduction, residuals/predictions blending between models
• Data-context aware: preparation statistics (mean, stdev) tracked between train/test splits
• All features, estimators, and their fits are composable, storable, and pluggable
• Alpha-stage library at time of writing; API subject to change

Connections: Scikit Learn · Github · Machine Learning · Feature Engineering · Model Selection

Source: https://github.com/kvh/ramp