Using the SageMaker training toolkit with scikit-learn
In this example, we're going to build a custom Python container with the SageMaker Training Toolkit. We'll use it to train a
scikit-learn model on the Boston Housing dataset, using Script Mode and the
We need three building blocks:
- The training script: Thanks to Script Mode, we can use exactly the same code as in the
Scikit-Learnexample from Chapter 7, Extending Machine Learning Services with Built-in Frameworks.
- We need a
Dockerfileand Docker commands to build our custom container.
- We also need a
SKLearnestimator configured to use our custom container.
Let's take care of the container:
Dockerfilecan get quite complicated. No need for that here! We start from the official Python 3.7 image available on Docker Hub (https://hub.docker.com/_/python). We install scikit-learn,
joblib, and the SageMaker training toolkit:
FROM python:3.7 RUN pip3 install...