We added:
- TCN Lightning module which can be driven through yaml configs and tracks everything to tensorboard
- Data augmentation methods
- Advanced financial CV methods
- FracDiff
- Time features
- Ability to pass a DF directly to most classes
- __init__ files so importing is less of a pain
We added:
- started to add binance data
- made imbalanced bars working
- sampling weights and downsampling based on that
- mini series creation and merging
- feature generation with TSFresh
- feature selection with TSFresh and Boruta
- working docs
In this very first semi-useful release we have:
- base bars (tested)
- imbalanced bars (un-tested, largely finished)
- labelling (paralellized, tested)
- snakemake pipelines for bar creation and labelling using dynamic YAML based runner