Datasets, DataFrames and RDDs¶
Many may have been asking yourself why they should be using Datasets rather than the foundation of all Spark - RDDs using case classes.
This document collects advantages of
RDD[CaseClass] to answer https://twitter.com/danosipov/status/704421546203308033[the question Dan has asked on twitter]:
"In #Spark, what is the advantage of a DataSet over an RDD[CaseClass]?"
Saving to or Writing from Data Sources¶
With Dataset API, loading data from a data source or saving it to one is as simple as using <
Accessing Fields / Columns¶
select columns in a datasets without worrying about the positions of the columns.
In RDD, you have to do an additional hop over a case class and access fields by name.