Discovering Amazon SageMaker Autopilot
Added to Amazon SageMaker in late 2019, Amazon SageMaker Autopilot is an AutoML capability that takes care of all machine learning steps for you. You only need to upload a columnar dataset to an Amazon S3 bucket, and define the column you want the model to learn (the target attribute). Then, you simply launch an Autopilot job, with either a few clicks in the SageMaker Studio GUI, or a couple of lines of code with the SageMaker SDK.
The simplicity of SageMaker Autopilot doesn't come at the expense of transparency and control. You can see how your models are built, and you can keep experimenting to refine results. In that respect, SageMaker Autopilot should appeal to new and seasoned practitioners alike.
In this section, you'll learn about the different steps of a SageMaker Autopilot job, and how they contribute to delivering high-quality models:
- Analyzing data
- Feature engineering
- Model tuning