sean_owen
Databricks Employee
Databricks Employee

You do not have to cache anything to make it work. You would decide that based on whether you want to spend memory/storage to avoid recomputing the DataFrame, like when you may use it in multiple operations afterwards.