This library allows for quick and easy benchmarking of a host of machine learning algorithms for supervised learning on general datasets.
Before you get into feature generation, extraction, or normalization, you can run this Python-based command line tool on your data to understand which algorithms give you the best initial performance. If you get similar performance from multiple algorithms, add noisy features and/or Gaussian noise to your features to check for tolerance to overfitting.
Works for both classification and regression problems. Examples on classical datasets (e.g. iris) are included.
For more information or requests for collaboration, please contact me at gaurav@sevenbridges.com.