Add autopilot customer churn notebook #1092
Conversation
Similar to the XGBoost customer churn notebook, but using autopilot instead.
|
Check out this pull request on You'll be able to see Jupyter notebook diff and discuss changes. Powered by ReviewNB. |
|
I would restrict the MaxCandidates to be less that 20 with explicit instruction on how to increase. https://docs.aws.amazon.com/sagemaker/latest/APIReference/API_AutoMLJobCompletionCriteria.html#sagemaker-Type-AutoMLJobCompletionCriteria-MaxCandidates This would imply that we may not have 96% accuracy as mentioned in the later section of the doc. |
Make sense. |
Will modify and run a couple of times to make sure I have a correct ballpark value in the text for the accuracy. |
So that the automl job completes within a few minutes we now limit the max candidates to 20. We've also added a note regarding the nature of the task, BinaryClassification, and how AutoPilot is able to auto detect it.
|
lgtm! |
Similar to the XGBoost customer churn notebook, but using autopilot
instead.
Issue #, if available:
Description of changes:
To support Aurora ML blog (https://aws.amazon.com/blogs/aws/new-for-amazon-aurora-use-machine-learning-directly-from-your-databases/) it is necessary to have a Customer Churn Autopilot example.
The blog currently references: https://github.com/awslabs/amazon-sagemaker-examples/blob/master/introduction_to_applying_machine_learning/xgboost_customer_churn/xgboost_customer_churn.ipynb but the resulting endpoint from this notebook is not usable from Aurora, since it does not include all the pre/post processing code generated.
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.