About

Fito aims to be the SQLAlchemy of the data science workflow. That means, the goal of fito is to help you specify experiments, and store metadata attached to them.

It then provides tools to help you on your day to day data science flow issues, so you can just focus on what really adds value.

What fito is not

Even though fito allows you to specify depencies between Operations, it does not pretend to be a pipelining technology like Luigi , Airflow or Pinball.

Also, fito does not pretend to provide a way to scale up you code, like what Spark does. You can just use fito over spark.

Neither fito pretends to implement efficient key value stores.

Fito just focuses on the ORM part of it. So if you like some key value tech, or you want to scale up with another tech, you can still use fito. You’ve just have to fitoize it by building a fito layer on top of it.

For example, fito.data_store.mongo.MongoHashMap provides the interface fito expects from a key value store with a mongo db backend. Same thing with a file system on fito.data_store.file.FileDataStore and using just a python dictionary on fito.data_store.dict.DictDataStore