Hi Luke Please check this issue once and let us know if merging HLL for uniques has been stable for you guys as we recently saw a huge discrepancy and wanted to check with you guys if we should totally avoid this option for now or we can bank upon it.
Thanks On Tue, Jan 12, 2016 at 9:04 AM, Prashant Prakash (JIRA) <[email protected]> wrote: > Prashant Prakash created KYLIN-1304: > --------------------------------------- > > Summary: Incorrect count estimate upon > HyperLogLogPlusCounter.merge operation. > Key: KYLIN-1304 > URL: https://issues.apache.org/jira/browse/KYLIN-1304 > Project: Kylin > Issue Type: Bug > Affects Versions: v1.1.1 > Reporter: Prashant Prakash > > > We are using hllc12 for evaluating uniques for a measure in a table > partitioned over date. > The uniques estimate for individual dates 2016-01-07, 2016-01-08, > 2016-01-09 are 93,728,324, 90,982,364, 45,485,278 respectively. > But the uniques across days, which is calculated through > HyperLogLogPlusCounter.merge operation gives a incorrect value 67,980,576. > > > > > -- > This message was sent by Atlassian JIRA > (v6.3.4#6332) >
