a CLI for (inter)active learning & data mining

key features

• simple syntax, build entire pipelines in a single line of code
• dynamic loading of your favourite libraries (you can use what you have access to)
• create a model from scratch, or load an existing one
• interactive user input via CLI

features to come

• more sampling strategies
• load/save models from MLFlow

Example with a simple logitic regression on the titanic dataset

Documentation

active_flow uses python fire internally, so you can call eg.

python -m aflow.active --help

Functional definitions