Skip to main content

Scripts in .ipynb files

You can use Jupyter notebook files as fal scripts.

init_fal magic command​

In order to access fal functions such as ref and write_to_model from within notebook runtime, you can use the init_fal magic command.

First, you need to import init_fal within a Python cell:

from faldbt.magics import init_fal

Now, you can initialize fal in your notebook:

%init_fal project_dir=project_dir profiles_dir=profiles_dir default_model_name=my_model

init_fal requires three inline arguments:

  • project_dir: path to the dbt project directory
  • profiles_dir: path to the dbt profiles directory
  • default_model_name: the model name that will be used in write_to_model, applies only in notebook runtime.

Once executed, you can use fal functions in your notebook's Python cells:

my_df = ref('some_model')

# We made some predictions and stored them in `my_predictions`

write_to_model(my_predictions)

Note that the default_model_name is only active during notebook runtime. When the script is run with fal run or fal flow run, fal will determine the model to write to according to the relevant schema.yml file. In fal runtime, the init_fal line is ignored.

You can specify a .ipynb file the same way as a regular Python file:

models:
  - name: some_model
    meta:
      fal:
        scripts:
            - my_notebook.ipynb