No need; To use DISTINCT COUNT, you only need define one measurement on the field, with the “hll” as data type;
On 4/15/15, 1:46 PM, "Shashank Prabhakara" <[email protected]> wrote: >Hi, >Related question: >In order to use DISTINCT COUNT, would it be correct to(from what I can >make >in the sample cube) to >1. A DISTINCT COUNT on a field >2. Define a measure with dependant_measure_ref on the first measure > >Thanks in advance, >Shashank > >On Tue, Apr 14, 2015 at 10:16 AM, Shi, Shaofeng <[email protected]> wrote: > >> Hi Sun, >> >> Good question; These two are advanced functions that still under >>internal >> evaluation. User couldn¹t create such measure from the web UI. To avoid >> such confusing, I will remove them from the sample cube (as the sample >> should be simple and easy to understand). >> >> Thanks for the feedback; >> >> Shaofeng, Shi >> >> >> On 4/14/15, 10:17 AM, "[email protected]" <[email protected]> wrote: >> >> >Hi, all >> > >> >Two questions : >> >1# The sample cube generated one measure listed as below. Notice that >> >the return type goes to bigint, however, if we create a cube for our >>own >> > and define such measure of count_distince expression, the return >>type >> >seems to be hllc(int) and no way to be bigint. How could that be >> >generated in the sample cube ? >> > >> > 6 SELLER_CNT_LONG COUNT_DISTINCT column SELLER_ID bigint >> > >> >2# According to the tutorial guide, Only accept single column in param >> >value with "Column" type. However, the sample cube ui listed as below >> >just show that the measure >> > contains 2 param value as LSTG_FORMAT_NAME and SELLER_ID. Quite >>weird >> >with the issue. >> > >> > 7 SELLER_FORMAT_CNT COUNT_DISTINCT column >>LSTG_FORMAT_NAME,SELLER_ID >> >hllc(10) >> > >> >AND, the tutorial listed here seems to be for older release versions of >> >kylin, maybe should go for an upgrade : >> > >> >>https://github.com/apache/incubator-kylin/blob/master/docs/Tutorial/Kylin >>% >> >20Cube%20Creation%20Tutorial.md >> > >> >Thanks for your attention, >> >Sun. >> > >> > >> > >> >[email protected] >> >>
