tree: ab7cd3c1763a31fa9d63cb28800ad05f4ec41332 [path history] [tgz]
  1. notebooks/
  2. dag.png
  3. notebook.ipynb
  4. README.md
  5. requirements.txt
  6. run.py
examples/contrib/README.md

Import Community Dataflows

In this example, we show you how to import and use dataflows from the Hamilton Dataflow Hub. You can use them either directly or pull and edit a local copy.

Setup

For the purpose of this example, we will create a virtual environment with hamilton, the hamilton contrib module, and the requirements for the dataflow we'll import.

  1. python -m venv ./venv
  2. . venv/bin/activate (on MacOS / Linux) or . venv/bin/Scripts (Windows)
  3. pip install -r requirements.txt

3 ways to import

There are 3 main ways to use community dataflows: static installation, dynamic installation, and local copy (see documentation). We present each of them in this example:

1. Static installation

The script run.py uses the direct import from hamilton.contrib.user.zilto import xgboost_optuna. It's as simple as that! (but first pip install sf-hamilton-contrib --upgrade)

2. Dynamic installation

The first part of the notebook notebook.ipynb imports the same dataflow via xgboost_optuna = hamilton.dataflows.import_module("xgboost_optuna", "zilto"). This will download and cache the module in your local directory {USER_PATH}/.hamilton.

3. Local copy

After completing the dynamic installation, the second part of the notebook includes hamilton.dataflows.copy(xgboost_optuna, destination_path="./my_local_path") will create a local copy at the desire location. Then, you'll be able to do from my_local_path import xgboost_optuna.

Contribute your own dataflow!

You can find more information on how to contribute in the contrib module's README