Creating a multi-step workflow: Model Engineering
The MLflow Projects module can be used as a way to run a multi-step workflow. All steps can be coordinated though a single Python program that passes results from previous steps to the following.
In this exercise, you will begin creating a multi-step workflow to manage the Model Engineering and Model Evaluation steps of the ML lifecycle. You will use the run() method from the MLflow Projects module for the model_engineering entry point and pass parameters used as hyperparameters for model training. You will also capture the output of the run_id and set it to a variable so that it can be passed to the model_evaluation step of the workflow as a parameter.
The MLproject created in the previous step is available in the IPython Shell using print(MLproject). The MLflow module is imported.
This exercise is part of the course
Introduction to MLflow
Exercise instructions
- Assign the
run()method from MLflow Projects module to a variable calledmodel_engineering. - Set the entry point argument to
"model_engineering". - Set parameters for training the model.
"n_jobs"to2and"fit_intercept"toFalse. - Set the
run_idattribute ofmodel_engineeringto a variable calledmodel_engineering_run_id.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
# Set run method to model_engineering
____ = ____.____.____(
uri='./',
# Set entry point to model_engineering
____='____',
experiment_name='Insurance',
# Set the parameters for n_jobs and fit_intercept
parameters={
'____': ____,
'____': ____
},
env_manager='local'
)
# Set Run ID of model training to be passed to Model Evaluation step
____ = ____.____
print(model_engineering_run_id)