Watch this video on YouTube
Your team is optimizing Spark performance for a machine learning project. What technique involves precomputing and storing intermediate results to reduce computational overhead during subsequent tasks?