sagemaker_sklearn_iris_training_job_and_endpoint_deployment.py

1import sagemaker
2from sagemaker.sklearn.estimator import SKLearn
3
4# Initialize the SageMaker session and get the execution role
5sagemaker_session = sagemaker.Session()
6role = sagemaker.get_execution_role()
7
8# Define the Scikit-learn estimator
9# Note: 'entry_point' should be a local script file (e.g., 'train.py') 
10# containing your training logic.
11sklearn_estimator = SKLearn(
12    entry_point='train.py',
13    role=role,
14    instance_count=1,
15    instance_type='ml.m5.large',
16    framework_version='1.2-1',
17    py_version='py3',
18    sagemaker_session=sagemaker_session
19)
20
21# Start the training job
22# 'train_data_uri' should be the S3 path to your training dataset
23sklearn_estimator.fit({'train': 's3://<your-bucket>/path/to/train/data'})
24
25# Deploy the model to an endpoint
26predictor = sklearn_estimator.deploy(
27    instance_count=1, 
28    instance_type='ml.m5.large'
29)
30
31# Predict using the endpoint
32# predictions = predictor.predict(data)
33
34# Delete the endpoint when finished
35# predictor.delete_endpoint()