This template is a good starter for making a new recipe. We recommend modifying this README to include information about the recipe you are creating. You should include a summary of your recipe as well as some examples of pipeline changes.
Describe your recipe here. We recommend including a gif or screenshot of the final output if applicable.
There is no need to modify the directions in the next section.
-
Ensure you have the following DataRobot feature flags turned on:
- INSERT REQUIRED FLAGS HERE
-
Create a new python virtual environment with python >= 3.9.
-
Install
kedro
, create a new kedro project from this template andcd
to the newly created directory. Choose a project name that is likely to be unique - DataRobot requires registered model names to be unique for an organization. You can change it later if necessary by editingparameters.yml
.pip install kedro
kedro new --name=your_project_name --starter=https://github_url_of_your_recipe_here --checkout main
cd your_project_name
-
Install requirements for this template:
pip install -r requirements.txt
-
Populate the following credentials in
conf/local/credentials.yml
:datarobot: endpoint: <your endpoint> # e.g. https://your_subdomain.datarobot.com/api/v2 api_token: <your api token>
-
Run the pipeline:
kedro run
. Start exploring the pipeline using the kedro GUI:kedro viz run
The following files govern pipeline execution. In general, you will not need to modify any other boilerplate files as you customize the pipeline.:
conf/base/parameters.yml
: pipeline configuration options and hyperparametersconf/local/credentials.yml
: API tokens and other secretsconf/base/catalog.yml
: file storage locations that can be used as node inputs or outputs, including locations of supporting assets to build DR custom models, execution environmentssrc/your_project_name/pipelines/*/nodes.py
: function definitions for the pipeline nodessrc/your_project_name/pipelines/*/pipeline.py
: node names, inputs and outputssrc/datarobotx/idp
: directory contains function definitions for for reusable idempotent DR nodesinclude/your_project_name
: directory contains raw assets and templates used by the pipeline
For a deeper orientation to kedro principles and project structure visit the Kedro documentation.
We recommend including some examples of changes users can make to this recipe in order to modify or extend its functionality. These changes could be in the parameters.yaml, the catalog.yaml file or in the pipeline itself.