SageMaker Canvas: specification of the predicted results csv path and name in S3

In Amazon SageMaker Canvas, when you generate predictions, the results are automatically stored in Amazon S3. By default, SageMaker Canvas creates a Canvas/ folder inside a bucket to store application data, including model artifacts, batch predictions, datasets, and logs.

During the initial setup of SageMaker Canvas, you have the option to configure where Canvas stores this data:

1. You can select "System managed" which uses the default SageMaker-created bucket following the pattern s3://sagemaker-{Region}-{your-account-id}

2. Alternatively, you can select "Custom S3" to specify your own Amazon S3 bucket as the storage location by entering the Amazon S3 URI

Additionally, you can optionally specify a KMS key for encrypting Canvas artifacts stored at the specified location.

However, the documentation doesn't indicate that you can specify the exact path or filename for individual prediction results within that bucket. The system appears to manage the specific file paths and naming conventions for prediction outputs automatically within the configured storage location.

When exporting data from Canvas data flows to Amazon S3, you do have more control - you can navigate to your data flow, choose the node to export, and specify the S3 location. But this is for exporting data flows rather than controlling where prediction results are saved.
Sources
Getting started with using Amazon SageMaker Canvas - Amazon SageMaker AI
Export data - Amazon SageMaker AI