Persisting intermediate feature DataFrames helps when:
Select an answer to reveal the explanation.
Short Explanation
persist/cache when multiple actions would recompute a heavy feature lineage.
Full Explanation
Spark lineage recomputation is a common footgun.