Data preparation involves data cleaning and feature engineering. It is the most time-consuming part of a machine learning project. Amazon ML offers powerful features to transform and slice the data. In this chapter, we will create the datasources that Amazon ML requires to train and select models. Creating a datasource involves three steps:
- Making the dataset available on AWS S3.
- Informing Amazon ML about the nature of the data using the schema.
- Transforming the initial dataset with recipes for feature engineering.
In a second part, we will extend Amazon ML data modification capabilities in order to carry out powerful feature engineering and data cleansing by using Amazon SQL service Athena. Athena is a serverless SQL-based query service perfectly suited for data munging in a predictive analytics context.