On Monday, March 2, 2015 at 4:23:19 PM UTC+2, Patrick Mast wrote:
>
>
> I do not fully understand why you put the week in the relationship type.
>

Speaking of 20+ millions of trips around 3+ thousands of locations, where 
20% of locations take 80% of stops (according to Pareto rule, but from my 
practice, both Pareto and Murphy were optimists; reality often gives us a 
more harsh form of power distribution, much like 9% vs 91% - but anyway) we 
end up with 600 "hot" locations taking 16 millions of stops - ~27 k stops 
per "hot" location. Splitting 27k relationships into "weekly" slices of 
size ~300 where week is coded in relationship type gives you ability to 
scan just ~300 relationships for any given week instead of scanning the 
whole 27k relationships each time. Plain simple. Maybe you just don't 
bother.
 

> As for your query: it does not prevent circles. Try to find a route from 
> Hamburg to Munich and you will get:
>
> 1. Hamburg --> Bremen --> Munich
> 2. Hamburg --> Hannover --> Munich
> 3. Hamburg --> Hannover --> Munich --> Frankfurt --> Munich
> 4. Hamburg --> Hannover --> Munich --> Frankfurt --> Munich --> Hannover 
> --> Hamburg --> Bremen --> Munich
>
> The last two results contain circles.
>

Hmm, my fault, sorry. I remember that it will avoid running *multiple* 
circles but won't avoid output of just a single circle. Sorry for that.

-- 
You received this message because you are subscribed to the Google Groups 
"Neo4j" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to