Individual pipelines are copied from vbflow into pipeline type files, such as gradient boosters going to gbr.py, in vb_django/vb_django/app.
Structure the class in the same manner as the existing examples, elasticnet.py and gbr.py.
Inherit from BaseEstimator, TransformerMixin, BaseHelper
Have a name, ptype, description, hyper_parameters and metrics class attributes
Initialize the class, also hitting super().init(pid)
Have a hyper_parameter validation function
Have a get pipeline function
Import the new pipeline class into vb_django.task_controller.py
Add the ptype of the new pipeline to the pipelines dict, starting at line 25 in task_controller.py
Add new conditional block in DaskTasks.execute_task, after line 92.
Class will be initialized
Hyper-parameters set
Call the fit function
Finally, call the save function
Any other additional functions that are necessary can be added here, such as the setup for cvpipe.
Some pipelines may not fall under this pattern, and will need to be more individually designed. But the sequential function calls should try to be maintained to help with code readability, troubleshooting and logging.