churn_prediction

This example is an end-to-end Bedrock tutorial on AWS. It demonstrates how to use Bedrock to preprocess data, train a model and create a HTTP endpoint for serving. This example complements Bedrock documentation.

Data exploration and Model prototyping

See notebook

Goals

At the end of the tutorial, the user will be able to

• set up a Bedrock training pipeline
• monitor the training
• deploy a model endpoint in HTTPS
• query the endpoint
• monitor the endpoint API metrics

Run on Bedrock

Just follow the Bedrock quickstart guide. We have already prepared the data on S3 so that you do not have to upload any.

Test your endpoint

After deploying your model as an endpoint, you can test it with the query below

curl -X POST \
  <MODEL_ENDPOINT_URL> \
  -H 'Content-Type: application/json' \
  -H 'X-Bedrock-Api-Token: <MODEL_ENDPOINT_TOKEN>' \
  -d '{"State": "ME", "Area_Code": 408, "Intl_Plan": 1, "VMail_Plan": 1, "VMail_Message": 21, "CustServ_Calls": 4, "Day_Mins": 156.5, "Day_Calls": 122, "Eve_Mins": 209.2, "Eve_Calls": 125, "Night_Mins": 158.7, "Night_Calls": 81, "Intl_Mins": 11.1, "Intl_Calls": 3}'