Exporting data for ML training
SageMaker Data Wrangler supports the following export options: Save to S3, Pipeline, Python Code, and Feature Store. The data transformations we have applied so far are not really applied to the data yet. The transformation steps need to be executed to get the final transformed data. When we export our flow file with the preceding options, SageMaker Data Wrangler automatically generates code and notebooks to guide you through the execution process so that we do not have to write any code, but it leaves flexibility for us to customize the code.
The four export options satisfy many use cases. Save to S3 is an obvious one and offers lots of flexibility. If you would like to get the transformed data in an S3 bucket so that you can train an ML model in Amazon SageMaker, you can also download it locally from S3 and import it to other tools if you need to. The Pipeline option creates a SageMaker pipeline that can easily be called a repeatable workflow. Such...
