Hi Mark,

Thanks once again for your advice. After playing around with the 
transformer the one thing that seemed to make a big difference was 
when I changed the "consider self" option from what I originally 
set, "no", to "yes". I was trying to find the closest polygon 
(candidate) to a line (base)and didn't think that this option would 
make a difference given the base and candidate were two discrete 
data sets. 

However, it seemed to as to process a set of 200 base records 
against a set of around 2000 candidate records went from over a day 
down to around 10 or so minutes.

Hugh.


--- In [email protected], "mark2atsafe" <[EMAIL PROTECTED]> wrote:
>
> Hi Hugh,
> I can't think of any obvious method - but I asked our development 
team
> and will file an enhancement request for a "Candidates First" 
option -
> in the same way as the Clipper has Bases First. That could help 
speed
> things up and reduce memory usage I'd think.
> 
> Besides that - the manual states... "After all features have been
> received, the factory then looks at each BASE feature and finds the
> closest CANDIDATE feature within <maxDistance> distance of the BASE
> feature."
> 
> ...so there are (candidate * base) comparisons going on and there
> isn't really anything you can do except cut down on the number of
> candidates by preprocessing the data.
> 
> For example, if your max distance is 100m then create a 100m buffer
> around the bases, overlay the candidates to find which are within 
this
> buffer and use them as the neighborFinder candidates.
> 
> Of course, the buffer/overlay would take some time - perhaps more 
than
> you would gain in the neighborfinder - and there's plenty of proof
> that our developers aren't dumb and would've thought to add that in
> automatically, but it's the only thing I can think of at this 
point.
> 
> Besides that - if it's a complex workspace with other 
transformers -
> be really sure that it is the NeighborFinder which is slow and not
> another part. There's a whole fmepedia category of performance 
related
> tips at... http://www.fmepedia.com/index.php/FME_Performance_Tuning
> 
> Hope this helps
> 
> Mark
> 
> Mark Ireland, Senior Product Specialist
> Safe Software Inc. Surrey, BC, CANADA
> [EMAIL PROTECTED] http://www.safe.com
> Solutions for Spatial Data Translation, Distribution and Access
> 
> --- In [email protected], "hughblackstock" <hughblackstock@> 
wrote:
> >
> > Hi There,
> > 
> > Do you have any advice on how to increase the speed that a 
> > neighborfinder transformer works? 
> > 
> > I'd like to find the closest polygon (suburb) to a line segment 
(road) 
> > but this seemingly simple task seems to take a long time to 
process 
> > even when I have a very small number of records.
> > 
> > Thanks,
> > 
> > Hugh.
> >
>








For insights into what's up at Safe Software and what's on the development 
horizon, visit Safe's blog at spatial-etl.blogspot.com.

Safe Software has also made slides available that outline enhancements planned 
for FME 2007. The slides are from the "Road Ahead" presentation given on Day 2 
of the FME Worldwide Users Conference. To view these slides, visit 
www.safe.com/2006uc.

 
Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/fme/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/fme/join
    (Yahoo! ID required)

<*> To change settings via email:
    mailto:[EMAIL PROTECTED] 
    mailto:[EMAIL PROTECTED]

<*> To unsubscribe from this group, send an email to:
    [EMAIL PROTECTED]

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/
 



Reply via email to