To work with SageMaker, we need to instantiate a Jupyter Notebook:
- In the AWS Management Console, search for SageMaker and press Enter.
- We can create a notebook from the SageMaker dashboard. Click on Create notebook instance.
- Give the instance a name, such as iiot-book-notebook, and set the instance type to be ml.t2.medium, as shown in the following screenshot:
SageMaker Notebook creation
- Amazon SageMaker will need a role to launch and access resources in the account. To simplify this, select Create a new role.
- After that, select the Specific S3 buckets option.
- Introduce the bucket name that you used in step 5 of the Downloading a dataset on S3 section:
Create role for SageMaker
- Clicking on Create role will create a role that contains the AmazonSageMakerFullAccess IAM managed policy.
- We now need to add the AmazonEC2ContainerRegistryFullAccess role. Click on the role that we created in the previous step and attach the appropriate policy. This is shown in the following screenshot:
Adding the AmazonSageMakerFullAccess policy to the role
- Select the No VPC option. Doing this will make it easier to configure access to the Amazon S3 bucket.
- Select No Custom Encryption.
- Click Create notebook instance and wait until the instance's status is InService.
- We are now ready to open the notebook for our code. Click on the Open button shown in the following screenshot:
Working with SageMaker Notebook
The other items in the menu on the left allow us to check the status of the training, create models, create endpoints, and see logs.