The sagemaker R package provides a simplified interface to the AWS Sagemaker API by:
data.frames and tibbles
Check out the Get started guide for examples!
The R package hides the details for later, so you can get started ASAP.
Get Sagemaker endpoint predictions with no string parsing or REST API management. Directly use
predict on the Sagemaker model to get predictions that conform to the tidymodel standard.
Even download and load Sagemaker trained models into your R session:
Easily get tuning and training stats with
You can install sagemaker from GitHub with:
You will also need
awscli python packages. If you don’t have them, or aren’t sure, install with:
Next, you’ll need an AWS account. This is a complicated process and unfortunately I don’t have a good guide to share. Please reach out if you find one and I’ll post it here!
Once you have the account and user setup you’ll want to save your account access keys. This is a good guide to follow.
Lastly, you need a AWS Sagemaker Execution Role. The easiest way is to create a AWS Sagemaker Notebook Instance. Try this guide to get the notebook running. Once the notebook is deployed, inside the instance run:
Once you have that role, run this command locally to save it with your local AWS config: