ml-meta-wrapper
The project aims to develop a customized ML framework on top of existing libraries
It will be written in python3.4+. Libraries which will be heavily relied upon are:
Pre-processing
- sci-kit learn (for e.g. outlier detection and standardizing)
- scipy
- More to follow...
Visualization
- Seaborn
- Hypertools
Learning and evaluation
- sci-kit learn
- tensorflow/keras
- xgboost
- More to follow...
Intepretation
- LIME
Wrapper modules will be added with time. Albeit most machine learning tasks fit on a single laptop, I will consider writing some modules in Spark (perhaps best suited for ETL jobs) or DASK/BLAZE.