I am currently working on Spark 1.6 using Scala. I want to get the quantiles of an integer column. Unfortunately, Spark doesn't have any quantile fun in 1.6. However, I found that we have percentile_approx() in hive. Is there any significant difference between those two? or Can I just use Percentile_approx instead of quantiles?
Asked
Active
Viewed 113 times
1
Raj
- 11
- 3
-
1you cant use percentile instead of quantile there is [difference](http://stats.stackexchange.com/a/156787) Spark2 has api support for this For <2 please [see](http://stackoverflow.com/a/31437177/647053) – Ram Ghadiyaram Jan 02 '17 at 07:51
-
Thanks for pointing me to the appropriate article @Ram – Raj Jan 10 '17 at 22:31