The sagemaker R package provides a simplified interface to the AWS Sagemaker API by:
- adding sensible defaults so you can dive in quickly
- creating helper functions to streamline model analysis
- supporting
data.frames and tibbles
Check out the Get started guide for examples!
Compare the AWS Sagemaker API vs. the sagemaker R package.
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.
model <- sagemaker_hyperparameter_tuner(xgb, s3_split(train, validation))
pred <- predict(model, new_data)Even download and load Sagemaker trained models into your R session:
local_model <- sagemaker_load_model(model)
local_pred <- predict(local_model, new_data)Easily get tuning and training stats with sagemaker_training_job_logs
and sagemaker_tuning_job_logs.
You can install sagemaker from GitHub with:
# install.packages("devtools")
devtools::install_github("tmastny/sagemaker")You will also need boto3, sagemaker, and awscli python packages.
If you don’t have them, or aren’t sure, install with:
sagemaker::sagemaker_install()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:
from sagemaker import get_execution_role
role = get_execution_role()
print(role)Once you have that role, run this command locally to save it with your local AWS config:
sagemaker::sagemaker_save_execution_role(
"arn:aws:iam::[account_number]:role/service-role/[SageMakerExecutionRole]"
)