Which Greenplum component is primarily responsible for generating an efficient execution plan based on query cost?

A data architect is designing a Greenplum database for an e-commerce company handling large datasets with frequent joins on customer_id. To optimize performance and ensure even data distribution across segments, what should the architect do?

Which two statements are true regarding the ST_DWithin function in PostGIS?
(Choose two)

Which two steps help improve ML model accuracy when using MADlib?
(Choose two)

How does Greenplum handle parallelism during ML model training using MADlib?