Train Autopilot models using a subset of data, and reuse its recipes on the whole dataset

This notebook's CI test result for us-west-2 is as follows. CI test results in other regions can be found at the end of the notebook.

Amazon SageMaker Autopilot is an automated machine learning (commonly referred to as AutoML) solution for tabular datasets. You can use SageMaker Autopilot in different ways: on autopilot (without any human input) or with human guidance, without code through SageMaker Studio or scripted using the AWS SDKs. This notebook will use the AWS SDKs to create an Autopilot job on a subset of data, and apply the recipes on a whole dataset. Compared with running Autopilot job directly on the whole dataset, you save cost and build models more efficiently.

Setup Environment

Prepare Data

Setup Autopilot Jobs

Kick off an autopilot job using only a subset of the training data.

For baselining purpose, we also kick off an autopilot job using the whole training data. Note that the solution itself only needs the first autopilot job.

Retrieve best models from the Autopilot job

Transform the whole training dataset with the best transform model

Create a new training job

Using the hyperparameters from the best candidate and fit on a larger dataset

Create a model using SageMaker AutoML API & Batch Transform Predict test dataset

We will replace the original tuning model in the autopilot job with the newly fitted model

Last Step! Evaluate the model and compare it with the baseline

The end

Notebook CI Test Results

This notebook was tested in multiple regions. The test results are as follows, except for us-west-2 which is shown at the top of the notebook.