Last week we covered "where and how to fetch data" (collection and storage). Now we move to the stage of transforming raw data into a form the model can consume. This is transformation, commonly called ETL (Extract-Transform-Load). In the MLS-C01 exam, ETL is a core area that tests "which tool to choose, at what scale, and with what cost and operational burden."
Today we compare AWS's representative transformation tools: AWS Glue (serverless Spark) and Amazon EMR (managed Hadoop/Spark cluster), and develop the perspective to design large-scale preprocessing.
ETL is a compound of three letters.