It would be nice if we could use null value in term/terms filters and
facets as indicator we want to filter/facet on missing values
for example following filter will return docs with country == canada or
blank (bool of terms and missing filter)
terms:{
country:['CANADA', null]
}
While it could easily be done combining terms and missing filters it is a
nice shortcut
More importantly in facet_filter null will be acting as a first class
citizen allowing us to get stats using terms_stats facets for missing filed
values
But the truth is that I would not be considering the functionality above if
terms_stats facet would have been consistent in calculating stats for
"missing" and "other" just like terms facet does with counts but
unfortunately it completely disregards them. I was working around it by
using statistical facet in the same query and using it to calculate "other
- missing" as difference between its total and sum of the total of each
term produced by my terms_stats facets. Unfortunately it does not let me
distinguish between "missing" and "other" which is very important for us to
produce consistent and complete charts. If use facet_filter would give me
stats for null I could have combined terms_stats query on unfiltered terms
with the same facet with facet_filter for null value
It is ugly but so
far I have not come up with any approach to calculate "missing" and "other"
stats in terms_stats facet
--
You received this message because you are subscribed to the Google Groups "elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email to elasticsearch+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.