# Amazon SageMaker Feature Store

## Introduction to Feature Store
In `feature_store_introduction.ipynb` we demonstrate how to get started with Feature Store, create feature groups, and ingest data into them.

This notebook requires these data sets in `./data/`:

* `feature_store_introduction_customer.csv`
* `feature_store_introduction_orders.csv`

This notebook requires these images in `./images/`:

* `feature-store-policy.png`
* `feature_store_data_ingest.svg`.

## Feature Store Encryption with KMS key
In `feature_store_kms_encryption.ipynb` we demonstrate how to encrypt data in your online or offline store using KMS key and how to verify that your KMS key is being used for data encryption. 

This notebook requires these data sets in `./data/`:
* `feature_store_introduction_customer.csv`
* `feature_store_introduction_orders.csv`

This notebook requires these images in `./images/`:
* `cloud-trails.png`
* `s3-sse-enabled.png`

## Client-side Encryption using AWS Encryption SDK
In `feature_store_client_side_encryption.ipynb` we demonstrate how client-side encryption with SageMaker Feature Store is done using the AWS Encryption SDK library to encrypt your data prior to ingesting it into your Online or Offline Feature Store. We first demonstrate how to encrypt your data using the AWS Encryption SDK library, and then show how to use Amazon Athena to query for a subset of encrypted columns of features for model training.

This notebook requires this synthetic data set in `./data/`:
* `credit_card_approval_synthetic.csv`

## Securely store an image dataset in your Feature Store with KMS key
In `feature_store_securely_store_images.ipynb` we demonstrate how to securely store a dataset of images into your Feature Store using KMS key. 

## Securely store the output of an image or text classification labelling job from Amazon Ground Truth directly into Feature Store using a KMS key
In `feature_store_classification_job_to_ground_truth.ipynb`, we demonstrate how to pipe the output of an image or text classification labelling job from Amazon Ground Truth directly into Feature Store. 

## Fraud Detection with Feature Store
For an advanced example on how to use Feature Store for a Fraud Detection use-case, see [Fraud Detection with Feature Store](https://sagemaker-examples.readthedocs.io/en/latest/sagemaker-featurestore/sagemaker_featurestore_fraud_detection_python_sdk.html), and it's associated notebook, `sagemaker_featurestore_fraud_detection_python_sdk.ipynb`.

## Developer Guide
For detailed information about Feature Store, see the [Feature Store Developer Guide](https://docs.aws.amazon.com/sagemaker/latest/dg/feature-store.html).