topic Re: Tool Processing times in Spatial Statistics Questions

Tool Processing times

DanielJarvis — Thu, 14 Nov 2013 16:52:54 GMT

Hi all,

I was wondering if anyone knew of a tool or calculator of some sort available to estimate processing times. For instance, I am using the Multi-Distance Spatial Cluster Analysis (Ripleys K Function) tool on a point shapefile that is 121mb and it has been running for a day and a half and is still not completed. I have defined an irregular extent and am using 999 random permutations for confidence intervals. The computer I am using has an i5 processor, is 64bit, and has 6g of RAM, and I am using arc10.2. I would like to be able make estimations of processing times for tools like this. Is that possible?

Thanks for any help you can offer,

Dan

Re: Tool Processing times

MarkJanikas — Fri, 15 Nov 2013 15:33:43 GMT

Hi Dan,

While k-function has had some speed improvements over the last few releases, it is an inherently slow algorithm. How many point features do you have? The 999 permutations are causing the delay... but it shouldn't run out of memory... so at some point it should finish.

At 10.2.1 we have moved a portion of our core calculations to C++ to improve performance. k-function is set for this improvement in the next release [i.e. post 10.2.1]. We got 50 times the speed for some of our clustering algorithms... so perhaps k-function will see a dramatic improvement as well in the future. That being said, the nature of the algorithm will always be poor from a Big O perspective.

Re: Tool Processing times

DanielJarvis — Fri, 15 Nov 2013 15:53:37 GMT

Hi Mark,

Thanks for the reply. I have 2,023,089 records in my point shapefile, so I expected a long processing time. It's still running at this moment, though for some reason it went from pulling 1.2GB of ram down to ~68MB..

It's good to know that the next release will likely be faster. I'm assuming then that at the moment there is no simple way to estimate processing times?

Dan