"In some cases, customer wants to have the flexibility to bring custom data processing code to SageMaker Autopilot. For example, customer might have datasets with large number of independent variables. Customer would like to have a custom feature selection step to remove irrelevant variables first. The resulted smaller dataset is then used to launch SageMaker Autopilot job. Customer would also like to include both the custom processing code and models from SageMaker Autopilot for easy deployment—either on a real-time endpoint or for batch processing. We will demonstrate how to achieve this in this notebook. \n",
0 commit comments