Session based analysis

Hi,

I am researching possibilities for us to use Druid as an advanced analytics engine. I am a little bit stuck on session analysis problem.

Let’s say I’ll be gathering user actions. Most of the stats I’d like to get I’ve already got. What I’d like to do next is:

  • calculate average session time

  • calculate average time between sessions per user

  • calculate Behavior Flow statistics (or funnels or user paths or whatever name there is for such) per session - i saw some thetaSketch based responses but none of them seemed to have

For the reference currently I am using Superset as a query/analysis UI. What is also worth mentioning is that this is web based project, so user entering event is easy to get, leaving time though is not that easy to determine unless we assume something like “lack of events in last X minutes means the end of sessions”.

Which of those calculations are possible in druid alone? How can they be achieved? Should I look for other tools to answer questions above based on druid realtime data? I will be greatful for examples of queries and/or dataSchemas - google-ing for such answers is painful since MMORPG-s and Hearthstone Druids are a bit more popular (I didn’t think I’ll ever say it, but damn you computer games!).

Thanks!

“i saw some thetaSketch based responses but none of them seemed to have…”
I meant to write that none of the proposals seemed to have chronological aspects to it, that is event X occured in users session after event Y.

W dniu poniedziałek, 20 lutego 2017 17:21:31 UTC+1 użytkownik RG napisał:

Hi all,

Bumping this as I have the same questions. Specifically, can someone with first-hand experience in this area address this part:

“none of the [thetaSketch] proposals seemed to have chronological aspects to it, that is event X occurred in users session after event Y.”

?

TIA,

Mikhail