Databoost

This github repo is the source code for the blog post Boosting as a scheme for transfer learning.

You should not attempt to understand what is here before reading that post.

That done? Great. Here's what's in this repo:

1. A few datasets (e.g. santander) used for testing.
2. A notebook multi_scenario.ipynb which was used to actually run the simulations and generate the graphs/tables in the blog post.
3. Source code in [databoost/][databoost/].

The source code has two files, utils.py which just has utilities for summing ML models, and scenario.py which has loaders for the different datasets used in testing.

The actual boosting code is in the notebook.