How to create model from best hyperparameter tuning job in Sagemaker?

Question

I'm using AWS SageMaker to run hyperparameter tuning to optimize an XGBoost model. I'm able to do so using the example code below. My question is: How can I then take the best hyperparameter tuning job and create a model via code? I'm aware that in the AWS console I can see the best job and click on "Create model", but I want to be able to do so using Python code.

sess = sagemaker.Session()
s3_input_train = sagemaker.s3_input(s3_data='s3://{}/{}/train'.format(bucket, prefix), content_type='csv')
s3_input_validation = sagemaker.s3_input(s3_data='s3://{}/{}/valid/'.format(bucket, prefix), content_type='csv')
xgb_cont = get_image_uri(region, 'xgboost', repo_version='0.90-1')
xgb = sagemaker.estimator.Estimator(xgb_cont, role, train_instance_count=1, train_instance_type='ml.m4.4xlarge',
                                    output_path='s3://{}/{}/output'.format(bucket, prefix), sagemaker_session=sess)
xgb.set_hyperparameters(eval_metric='rmse', objective='reg:squarederror', num_round=100)
hyperparameter_ranges = {'eta': ContinuousParameter(0, 1), 'min_child_weight': ContinuousParameter(1, 10),
                        'alpha': ContinuousParameter(0, 2), 'max_depth': IntegerParameter(1, 10)}
tuner = HyperparameterTuner(xgb, 
                            objective_metric_name='validation:rmse', 
                            objective_type='Minimize',
                            hyperparameter_ranges=hyperparameter_ranges, 
                            max_jobs=20, max_parallel_jobs=10)
tuner.fit({'train': s3_input_train, 'validation': s3_input_validation})

Yes, as advised by Gaurav Bansal above you can call the deploy method on the HyperParameterTuner object. More info here: sagemaker.readthedocs.io/en/stable/api/training/… — Ajay Mahendru

Gaurav Bansal Gaurav Bansal · Accepted Answer · 2020-03-06T17:10:32

Apparently I can deploy the best model using the following code:

tuner.deploy(initial_instance_count=1, instance_type='ml.m4.xlarge')

How to create model from best hyperparameter tuning job in Sagemaker?

1 Answers