org.apache.spark.sql.catalyst.util
Statistics from the Greenwald-Khanna paper.
Statistics from the Greenwald-Khanna paper.
the sampled value
the minimum rank jump from the previous value's minimum rank
the maximum span of the rank.
The default value for the compression threshold.
The size of the head buffer.
The default value for the relative error (1%).
The default value for the relative error (1%). With this value, the best extreme percentiles that can be approximated are 1% and 99%.