[SPARK-2032] Add an RDD.samplePartitions method for partition-level sampling

• Type: Improvement
• Status: Resolved
• Priority: Minor
• Resolution: Won't Fix
• Component/s: Spark Core
• Assignee: Prashant Sharma
• Reporter: Matei Alexandru Zaharia
• Created: 05/Jun/14 08:12
• Updated: 07/Feb/20 17:23
• Resolved: 08/Oct/16 09:33

Description

This would allow us to sample a percent of the partitions and not have to materialize all of them. It's less uniform but much faster and may be useful for quickly exploring data.