Stats vs Analytics
Trying to make sense of all issues around this and not sure which way to go. Both Stats and Analytics component are missing some features I would need. Stats cannot limit or order facets for instance, and I'd like to see pivot support. On the other end Analytics doesn't support distribution at all, which is a must in my case. So, I guess what I'm trying to ask is whether I should look at extending Stats or Analytics? Which way is the community going for future releases? (Would share any extension, but that would be useless if done on the wrong component). Thanks, Steve - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
Re: Stats vs Analytics
Just to add more discussion to the mix, we're also building/using this at CareerBuilder: Percentiles for facets, pivot facets, and distributed pivot facets https://issues.apache.org/jira/browse/SOLR-3583 It is an extension to (distributed pivot) faceting that allows stats to be collected within the faceting component. We built it with the following needs: 1) Supports pivot faceting (stats at each level) 2) Supports distributed statistical operations If you look at slide 41 of this presentation, you'll get a really good feel for what this patch does: http://www.slideshare.net/treygrainger/building-a-real-time-big-data-analytics-platform-with-solr The primary focus initially was on calculating percentiles of numerical values in a distributed way (using bucketing similar to range faceting), but we are also in the process of adding distributed sum. Other distributable calculations are possible, we just haven't needed them yet so we haven't added them. -Trey On Tue, Feb 11, 2014 at 2:24 PM, Steve Molloy smol...@opentext.com wrote: Trying to make sense of all issues around this and not sure which way to go. Both Stats and Analytics component are missing some features I would need. Stats cannot limit or order facets for instance, and I'd like to see pivot support. On the other end Analytics doesn't support distribution at all, which is a must in my case. So, I guess what I'm trying to ask is whether I should look at extending Stats or Analytics? Which way is the community going for future releases? (Would share any extension, but that would be useless if done on the wrong component). Thanks, Steve - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
Re: Stats vs Analytics
Thanks, will look further into it, but at first glance this looks like it may make things a lot more simple for me. This said, I'm still curious about the stats vs analytics approach for future releases. Thanks, Steve On Feb 11, 2014, at 5:03 PM, Trey Grainger solrt...@gmail.com wrote: Just to add more discussion to the mix, we're also building/using this at CareerBuilder: Percentiles for facets, pivot facets, and distributed pivot facets https://issues.apache.org/jira/browse/SOLR-3583 It is an extension to (distributed pivot) faceting that allows stats to be collected within the faceting component. We built it with the following needs: 1) Supports pivot faceting (stats at each level) 2) Supports distributed statistical operations If you look at slide 41 of this presentation, you'll get a really good feel for what this patch does: http://www.slideshare.net/treygrainger/building-a-real-time-big-data-analytics-platform-with-solr The primary focus initially was on calculating percentiles of numerical values in a distributed way (using bucketing similar to range faceting), but we are also in the process of adding distributed sum. Other distributable calculations are possible, we just haven't needed them yet so we haven't added them. -Trey On Tue, Feb 11, 2014 at 2:24 PM, Steve Molloy smol...@opentext.com wrote: Trying to make sense of all issues around this and not sure which way to go. Both Stats and Analytics component are missing some features I would need. Stats cannot limit or order facets for instance, and I'd like to see pivot support. On the other end Analytics doesn't support distribution at all, which is a must in my case. So, I guess what I'm trying to ask is whether I should look at extending Stats or Analytics? Which way is the community going for future releases? (Would share any extension, but that would be useless if done on the wrong component). Thanks, Steve - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org - To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org